Spiking Neural Networks (SNNs) offer a promising energy-efficient computing paradigm owing to their event-driven properties and biologically inspired dynamics. Among various encoding schemes, Time-to-First-Spike (TTFS) is particularly notable for its extreme sparsity, utilizing a single spike per neuron to maximize energy efficiency. However, two significant challenges persist: effectively leveraging TTFS sparsity to minimize training costs on Graphics Processing Units (GPUs), and bridging the performance gap between TTFS-based SNNs and their rate-based counterparts. To address these issues, we propose a parallel training algorithm for accelerated execution and a novel decoding strategy for enhanced performance. Specifically, we derive both forward and backward propagation equations for parallelized TTFS SNNs, enabling precise calculation of first-spike timings and gradients. Furthermore, we analyze the limitations of existing output decoders and introduce a membrane potential–based decoder, complemented by an incremental time-step training strategy, to improve accuracy. Our approach achieves state-of-the-art accuracy for TTFS SNNs on several benchmarks, including MNIST (99.51%), Fashion-MNIST (93.14%), CIFAR-10 (95.06%), and CIFAR-100 (74.07%).

Understanding the structural dynamics of biomolecules is crucial for uncovering biological functions. As molecular dynamics (MD) simulation data becomes more available, deep generative models have been developed to synthesize realistic MD trajectories. However, existing methods produce fixed-length trajectories by jointly denoising high-dimensional spatiotemporal representations, which conflicts with MD’s frame-by-frame integration process and fails to capture time-dependent conformational diversity. Inspired by MD's sequential nature, we introduce a new probabilistic autoregressive (ProAR) framework for trajectory generation. ProAR uses a dual-network system that models each frame as a multivariate Gaussian distribution and employs an anti-drifting sampling strategy to reduce cumulative errors. This approach captures conformational uncertainty and time-coupled structural changes while allowing flexible generation of trajectories of arbitrary length. Experiments on ATLAS, a large-scale protein MD dataset, demonstrate that for long trajectory generation, our model achieves a 7.5% reduction in reconstruction RMSE and an average 25.8% improvement in conformation change accuracy compared to previous state-of-the-art methods. For conformation sampling task, it performs comparably to specialized time-independent models, providing a flexible and dependable alternative to standard MD simulations.

World-model-based reinforcement learning achieves high sample efficiency by learning from imagined rollouts. However, its success critically depends on the accuracy of the learned world model, which is prone to producing unrealistic or hallucinated rollouts when queried beyond its domain of competence. These flawed predictions can trap the agent in a vicious cycle: by misleading exploration toward implausible or uninformative regions, they degrade the quality of collected data, which in turn corrupts policy learning with inaccurate rollouts. To break this cycle, we introduce the notion of a knowledge boundary—the region within which the world model provides reliable predictions—and propose a unified framework that both identifies and leverages this boundary. Concretely, we approximate the boundary using model uncertainty, quantified via disagreement across an ensemble of lightweight predictors, which serves as a practical proxy. This uncertainty signal is used in two complementary ways: as an intrinsic reward to guide exploration toward under-explored yet learnable regions, and as a dynamic filter to exclude unreliable imagined rollouts from policy optimization. Extensive experiments across diverse benchmarks—including CARLA, DeepMind Control Suite, Atari, and MemoryMaze—demonstrate that our approach consistently outperforms prior state-of-the-art methods.

Visual grounding, the task of linking textual queries to specific regions within images, plays a pivotal role in vision-language integration. Existing methods typically rely on extensive task-specific annotations and fine-tuning, limiting their ability to generalize effectively to novel or out-of-distribution scenarios. To address these limitations, we introduce GroundingAgent, a novel agentic visual grounding framework that operates without any task-specific fine-tuning. GroundingAgent employs a structured, iterative reasoning mechanism that integrates pretrained open-vocabulary object detectors, multimodal large language models (MLLMs), and large language models (LLMs) to progressively refine candidate regions through joint semantic and spatial analyses. Remarkably, GroundingAgent achieves an average zero-shot grounding accuracy of 65.1% on widely-used benchmarks (RefCOCO, RefCOCO+, RefCOCOg), entirely without fine-tuning. Furthermore, by substituting MLLM-generated captions with the original query texts, the accuracy at the selection stage alone reaches approximately 90%, closely matching supervised performance and underscoring the critical role of LLM reasoning capabilities. GroundingAgent also offers strong interpretability, transparently illustrating each reasoning step, thus providing clear insights into its decision-making process.

Spiking neural networks (SNNs) offer a promising energy-efficient alternative to artificial neural networks, due to their event-driven spiking computation. However, some foundation SNN backbones (including Spikformer and SEW ResNet) suffer from non-spike computations (integer-float multiplications) caused by the structure of their residual connections. These non-spike computations increase SNNs' power consumption and make them unsuitable for deployment on mainstream neuromorphic hardware. In this paper, we analyze the spike-driven behavior of the residual connection methods in SNNs. We then present Spikingformer, a novel spiking transformer backbone that merges the MS Residual connection with Self-Attention in a biologically plausible way to address the non-spike computation challenge in Spikformer while maintaining global modeling capabilities. We evaluate Spikingformer across 13 datasets spanning large static images, neuromorphic data, and natural language tasks, and demonstrate the effectiveness and universality of Spikingformer, setting a vital benchmark for spiking neural networks. In addition, with the spike-driven features and global modeling capabilities, Spikingformer is expected to become a more efficient general-purpose SNN backbone towards energy-efficient artificial intelligence.

Large language models (LLMs) have demonstrated significant reasoning capabilities in scientific discovery but struggle to bridge the gap to physical execution in wet-labs.In these irreversible environments, probabilistic hallucinations are not merely incorrect; they can cause equipment damage or experimental failure.We propose BioProAgent, a neuro-symbolic framework that anchors probabilistic planning in a deterministic Finite State Machine (FSM).We introduce a State-Augmented Planning mechanism that enforces a rigorous Design-Verify-Rectify workflow, ensuring hardware compliance before execution.Furthermore, we address the context bottleneck inherent in complex device schemas by Semantic Symbol Grounding, reducing token consumption by ∼6× through symbolic abstraction.In the extended BioProBench benchmark, BioProAgent achieves 95.6% physical compliance (compared to 21.0% for ReAct), demonstrating that neuro-symbolic constraints are essential for reliable autonomy in irreversible physical environments.

With the rapid growth of video content on social media, video summarization has become a crucial task in multimedia processing. However, existing methods face challenges in capturing global dependencies in video content and accommodating multimodal user customization. Moreover, temporal proximity between video frames does not always correspond to semantic proximity. To tackle these challenges, we propose a novel Language-guided Graph Representation Learning Network (LGRLN) for video summarization. Specifically, we introduce a video graph generator that converts video frames into a structured graph to preserve temporal order and contextual dependencies. By constructing forward, backward and undirected graphs, the video graph generator effectively preserves the sequentiality and contextual relationships of video content. We designed an intra-graph relational reasoning module with a dual-threshold graph convolution mechanism, which distinguishes semantically relevant frames from irrelevant ones between nodes. Additionally, our proposed language-guided cross-modal embedding module generates video summaries with specific textual descriptions. We model the summary generation output as a mixture of Bernoulli distribution and solve it with the EM algorithm. Experimental results show that our method outperforms existing approaches across multiple benchmarks. Moreover, we proposed LGRLN reduces inference time and model parameters by 87.8% and 91.7%, respectively.

Recent advances in Neural Architecture Search (NAS) are essentially attributed to Performance Estimation (PE), i.e., a method aims to effectively estimate an architecture. Meanwhile, Kendall's $\tau$τ is well recognized as the principled evaluation criteria for PE strategies in the literature. We argue that Kendall's $\tau$τ is not the optimal solution. Through extensive experiments and theoretical analysis, we take the initiative to reveal the problem behind the Kendall's $\tau$τ and propose a novel criterion named Minimum Keeping Ratio (MKR), which is closely connected to the final performance of NAS. It allows us to compare different PE approaches in a unified perspective, and use effective ablation studies to verify common beliefs and key differences of PE strategies. Based on the findings from MKR, we are able to derive a simple NAS method by integrating different PE strategies with random sampling. Such a method shows very strong performance in efficiency and effectiveness through extensive experiments on different challenging benchmarks. In particular, our simple random sampling NAS finds the optimal architecture in NASbenchMacro, NASbench201, and NASbench301. It is also well generalized to different search spaces (MobileNet) and tasks (semantic segmentation), finding an architecture surpasses the previous state-of-the-art architectures by 4.25 mIoU under $ 600M$600M FLOPs on ADE20K. Codes are available at https://anonymous.4open.science/r/Anonymization11264.

Generalized Category Discovery (GCD) is a challenging task that aims to identify both seen and novel categories in unlabeled data. We argue that a clear margin between seen and novel class representations is essential for accurate recognition. However, existing methods often ignore this margin, mapping representations to prototypes without enforcing separation between seen and novel classes. This leads to a bias where seen samples are misclassified as novel. To address this issue, we propose DebiasGCD, a debiasing framework that enhances prototype separation through margin-aware learning. Unlike prior work that relies on static prototype learning and overlooks fine-grained representations, our method introduces Dynamic Prototype Debiasing (DPD) and Spatial-Aware Representation Distillation (SARD) to mitigate this bias. First, DPD dynamically enforces inter-prototype margins, improving class-specific feature learning and prototype discrimination. Meanwhile, SARD promotes local representation of spatial learning, supporting DPD to capture subtle details that further refine class-specific features. By synergizing these components, DebiasGCD significantly improves prototype discriminability, generating more reliable predictions for seen classes. Extensive experiments demonstrate that our approach effectively mitigates pseudo-labeling bias across datasets, especially on fine-grained ones, achieving +8.3% and +9.6% improvements on the ‘All’ classes in CUB and Stanford Cars, respectively.

Neural Radiance Field (NeRF) utilizes Artificial Neural Networks (ANNs) to map 3D points and their corresponding 2D viewing directions to colors and densities. However, this approach encounters several challenges due to the inherent characteristics of ANNs. Firstly, ANNs tend to extract smooth features across sampled points, which makes it difficult to accurately represent the non-smooth variations between object surfaces and the surrounding air. Secondly, ANNs compute densities and colors independently for each point, failing to account for the sequential dependencies among points along the same ray. To tackle these issues, Spiking Neural Networks (SNNs) are introduced, which are better at processing sequential information and non-smooth representations. In this work, we propose a novel hybrid NeRF framework called SAH-NeRF that combines ANNs with SNNs. By harnessing ANNs’ robust representation capabilities alongside SNNs’ strengths in handling non-smooth distributions, our method could significantly improve the performance of three ANN-based NeRFs, surpassing state-of-the-art methods including 3D Gaussian Splatting. Notably, our SAH-NeRF could meanwhile enhance novel view synthesis and reduce energy consumption.

Quantization and chroma downsampling are two primary operations that introduce distortions in the JPEG compression. However, most existing blind methods treat artifacts removal as a direct mapping from compressed images to clean ones. They fail to explicitly model the underlying degradation process or design targeted compensation mechanisms. As a result, these methods can only partially remove compression artifacts and struggle to generalize to diverse or unseen degradation scenarios. In this work, we present a novel perspective that formulates artifacts removal as an approximate inversion of the lossy steps in JPEG. Based on this view, we propose an Inverse JPEG Compression Network (IJCN), which aims to progressively compensate for quantization errors and color distortions. Specifically, we first design a Learnable Offset Guidance Module (LOGM) to approximate inverse quantization by modeling both intra-block and inter-block coefficient correlations for predicting rounding offsets. In addition, we propose a Quantization Table Guidance Module (QTGM) that leverages the quantization tables to guide the reconstruction network in mitigating color distortions. By modeling compensation mechanisms under the guidance of quantization tables, IJCN effectively eliminates artifacts across varying compression levels. Extensive experiments demonstrate that IJCN outperforms existing methods in both quantitative metrics and visual quality.

3D facial avatar reconstruction is a fundamental problem in computer vision and graphics with applications in digital humans, virtual reality, and telepresence. Recent neural radiance field (NeRF)-based methods have greatly improved fidelity, yet most remain subject-specific, requiring multi-view images with diverse expressions and extensive test-time finetuning, which limits their generalization to unseen identities. Achieving high-quality reconstruction from a single image is particularly challenging due to missing multi-view supervision and the need to balance efficiency with fidelity. To address these issues, we present NOFA++, a generalizable one-shot framework that reconstructs photorealistic 3D facial avatars from a single input. Our method leverages the generative prior of a pretrained 3D GAN in an encoder–generator pipeline to recover a canonical neural volume, and introduces a coarse-to-fine residual generation strategy to synthesize identity-specific details without per-subject optimization. We further design a deformation field conditioned on identity and expression parameters to model facial dynamics, enabling controllable reenactment from video or audio. Extensive experiments show that NOFA++ surpasses state-of-the-art baselines in both reconstruction fidelity and reenactment quality, while eliminating test-time finetuning and generalizing robustly across unseen subjects.

Cross-modal retrieval is essential for exploring semantic correlations between multimodal data. However, existing approaches face challenges in resolving semantic ambiguity and transferring knowledge with sparse sample generalization. To address these challenges, we propose a new Semantic-Decoupled and Knowledge-Shared Probabilistic Mapping Network (SKPMN). Specifically, the Semantic Decoupling and Distinction (SDD) module decomposes complex word-region relationships into relevance-driven representations. The Deep Probability Mapping (DPM) module introduces a paradigm shift by mapping multimodal features into probabilistic distributions, capturing the semantic similarities and the potential uncertainties that define sparse or ambiguous relationships. By combining the Attention Probabilistic Mapping (APM) module, the model can effectively transfer knowledge across similar samples while emphasizing critical distinctions, significantly enhancing generalization to sparse and ambiguous samples. Finally, the multi-grained alignment strategy establishes a novel integration of fine-grained patch-to-word alignment and coarse-grained global alignment. Experimental results show that SKPMN achieves superior retrieval accuracy across major benchmark datasets. Furthermore, we implement a channel resource allocation technique that allocates more transmission resources to semantically significant information. In resource-constrained environments, our approach leverages Joint Source-Channel Coding (JSCC) to enhance the efficiency of visual feature transmission.

Existing end-to-end methods for steering angle prediction (SAP) primarily rely on RGB imagery from conventional cameras as input; however, they suffer from limitations such as poor performance in low-light conditions and motion blur. Recently, event cameras have garnered attention as complementary to RGB imagery, providing advantages such as high dynamic range and low latency. Nevertheless, earlier SAP methods that integrate event and RGB data may not fully exploit the spatio-temporal characteristics of events, resulting in performance degradation in low-light scenarios affected by noise interference. To address this limitation, we present a novel spatiotemporal-aware event-RGB fusion method for SAP, referred to as SERF, which aims to enhance the accuracy of event-based SAP. Specifically, SERF introduces three key components: 1) An innovative multi-layer Interaction Module based on attention mechanisms to fuse the multi-frame data, enabling more fine-grained feature processing; 2) a dynamic spatiotemporal mask mechanism, focusing RGB’s attention on spatially proximate events while diminishing the influence of temporally distant events, thereby reducing the impact of noise; and 3) a Memory Module that utilizes learnable tokens to accumulate essential latent fusion features through dynamic feature consolidation. Extensive experiments conducted on a variety of real-world and simulated datasets demonstrate the superior performance of SERF compared to the state-of-the-art methods. The experiments also validate the advantages of SERF in terms of inference performance, meeting the real-time requirements for actual deployment.

In federated recommendation systems, model poisoning attacks aim to manipulate the gradient information of multiple target items sent back from local clients to the central server, with the goal of abnormally increasing their exposure across the system. Existing multi-target attack approaches directly manipulate multiple target items and apply a uniform attack strategy to all target items, which may lead to suboptimal promotion effectiveness. To address this issue, we introduce ProitMTA, a novel multi-target model poisoning attack framework that introduces proxy items and provides tailored attack strategies for target items. ProitMTA employs a three-stage process that balances the promotion of multiple target items while preserving recommendation quality. First,proxy item generationuses a Gaussian Mixture Model to create proxy items that represent diverse attack strategies. Second,proxy attack constructiondesigns customized gradient manipulation strategies for each proxy item. Finally,proxy-based target item attacktransfers these strategies to actual target items, enhancing their promotion while minimizing the negative impact on system performance. Through comprehensive experiments on multiple base federated recommendation frameworks and diverse real-world datasets, we demonstrate that ProitMTA outperforms existing attack methods, achieving higher success rates in target item promotion with minimal system-wide performance degradation. Our research highlights the vulnerability of federated recommendation systems when facing multi-target poisoning attacks and underscores the importance of researching effective defense mechanisms We have released our code athttps://github.com/zdy769243418/ProitMTA.

X-ray image based medical report generation achieves significant progress in recent years with the help of large language models, however, these models have not fully exploited the effective information in visual image regions, resulting in reports that are linguistically sound but insufficient in describing key diseases. In this paper, we propose a novel associative memory-enhanced X-ray report generation model that effectively mimics the process of professional doctors writing medical reports. It considers both the mining of global and local visual information and associates historical report information to better complete the writing of the current report. Specifically, given an X-ray image, we first utilize a classification model along with its activation maps to accomplish the mining of visual regions highly associated with diseases and the learning of disease query tokens. Then, we employ a visual Hopfield network to establish memory associations for disease-related tokens, and a report Hopfield network to retrieve report memory information. This process facilitates the generation of high-quality reports based on a large language model and achieves state-of-the-art performance on multiple benchmark datasets, including the IU X-ray, MIMIC-CXR, and Chexpert Plus. The source code and pre-trained models of this work have been released on https://github.com/Event-AHU/Medical_Image_Analysis.

Rendering complex dynamic scenes offers the advantage of observing and understanding the real world. However, existing Dynamic Scene Rendering (DSR) methods remain challenged by suboptimal reconstruction fidelity. These limitations stem from relying on a single, unified deformation model, which struggles to capture complex motions involving multiple sub-motions and abrupt geometric transitions. While temporal decomposition methods could alleviate such shortcomings, they introduce the additional challenge of ignoring motion correlations and increasing storage requirements. To address these issues, we introduce Keyframe-driven Predictable Gaussian Splatting (KPGS)-an efficient framework for high-fidelity complex dynamic scene rendering. First, we present a patch-wise HSV clustering for extracting keyframes. Second, a prediction network based on the Transformer is utilized to calculate the deformable Gaussians at discrete keyframe times via voxelization. Third, we propose an inter-frame deformation network and a mutual supervision between adjacent segments to maintain the temporal continuity. Extensive experiments on our newly built dataset (MotionGS), as well as public benchmarks HyperNeRF and Neu3D, demonstrate that KPGS could achieve a higher average view synthesis performance than SOTA approaches, while maintaining a balance between storage cost and performance. More details of the demo and dataset are available at KPGS Supplementary.

3D scene generation is in high demand across various domains, including virtual reality, gaming, and the film industry. Owing to the powerful generative capabilities of text-to-image diffusion models that provide reliable priors, creating 3D scenes using only text prompts has become viable, thereby significantly advancing research in text-driven 3D scene generation. Prevailing methods typically employ the diffusion model to generate an initial local image, followed by iteratively outpainting the local image to gradually generate scenes. Nevertheless, these outpainting-based approaches are prone to producing globally inconsistent results with low completeness, restricting their broader applications. To tackle these problems, we introduce HoloDreamer, a framework that begins by generating a high-definition panorama to holistically initialize the full scene, and leverages 3D Gaussian Splatting (3D-GS) for rapid 3D scene reconstruction, thereby facilitating the creation of view-consistent and fully enclosed 3D scenes. Specifically, we propose Stylized Equirectangular Panorama Generation, a pipeline that combines multiple diffusion models to enable stylized and detailed equirectangular panorama generation from complex text prompts. Subsequently, Enhanced Two-Stage Panorama Reconstruction is introduced, conducting a two-stage optimization of 3D-GS to inpaint the missing region and enhance the integrity of the scene. Comprehensive experiments demonstrated that our method outperforms prior works in terms of overall visual consistency and harmony, as well as reconstruction quality and rendering robustness when generating fully enclosed scenes.

Compared to frame-based methods, computational neuromorphic imaging using event cameras offers significant advantages, such as minimal motion blur, enhanced temporal resolution, and high dynamic range. The multi-view consistency of Neural Radiance Fields combined with the unique benefits of event cameras, has spurred recent research into reconstructing NeRF from data captured by moving event cameras. While showing impressive performance, existing methods rely on ideal conditions with the availability of uniform and high-quality event sequences and accurate camera poses, and mainly focus on the object level reconstruction, thus limiting their practical applications. In this work, we propose AE-NeRF to address the challenges of learning event-based NeRF from non-ideal conditions, including non-uniform event sequences, noisy poses, and various scales of scenes. Our method exploits the density of event streams and jointly learn a pose correction module with an event-based NeRF (e-NeRF) framework for robust 3D reconstruction from inaccurate camera poses. To generalize to larger scenes, we propose hierarchical event distillation with a proposal e-NeRF network and a vanilla e-NeRF network to resample and refine the reconstruction process. We further propose an event reconstruction loss and a temporal loss to improve the view consistency of the reconstructed scene. We established a comprehensive benchmark that includes large-scale scenes to simulate practical non-ideal conditions, incorporating both synthetic and challenging real-world event datasets. The experimental results show that our method achieves a new state-of-the-art in event-based 3D reconstruction.

Learning control policy from continuous action space by visual observations is a fundamental and challenging task in reinforcement learning (RL). An essential problem is how to accurately map the high-dimensional images to the optimal actions by the policy network. Traditional decision-making modules output actions solely based on the current observation, while the distributions of optimal actions are dependent on specific tasks and cannot be known priorly, which increases the learning difficulty. To make the learning easier, we analyze the action characteristics in several control tasks, and propose Reinforcement Learning with Residual Action (ResAct) to explicitly model the adjustments of actions based on the differences between adjacent observations, rather than learning actions directly from observations. The method just redefines the output of the policy network, and doesn’t introduce any prior assumption to constrain or simplify the vanilla control problem. Extensive experiments on DeepMind Control Suite and CARLA demonstrate that the method could improve different RL baselines significantly, and achieve state-of-the-art performance.

Large Language Models (LLMs) often suffer from catastrophic forgetting when learning multiple tasks sequentially, making continual learning (CL) essential for their dynamic deployment. Existing state-of-the-art (SOTA) methods, such as O-LoRA, typically focus on constructing orthogonality tasks to decouple parameter interdependence from various domains.In this paper, we reveal that building non-collision parameters is a more critical factor in addressing CL challenges. Our theoretical and experimental analyses demonstrate that non-collision parameters provide better task orthogonality, which is a sufficient but unnecessary condition. Furthermore, knowledge from multiple domains will be preserved in non-collision parameter subspaces, making it more difficult to forget previously seen data. Leveraging this insight, we propose Non-collision Low-Rank Adaptation (N-LoRA), a simple yet effective approach leveraging low collision rates to enhance CL in LLMs. Experimental results on multiple CL benchmarks indicate that N-LoRA achieves superior performance (+2.9%), higher task orthogonality (×4.1times), and lower parameter collision (×58.1times) than SOTA methods.

Large Language Models (LLMs) have demonstrated powerful performance in sequential recommendation due to their robust language modeling and comprehension capabilities. In such paradigms, the item texts of interaction sequences are formulated as sentences and LLMs are utilized to learn language representations or directly generate target item texts by incorporating instructions. Despite their promise, these methods solely focus on modeling the mapping from sequential texts to target items, neglecting the relationship between the items in an interaction sequence. This results in a failure to learn the transition patterns between items, which reflect the dynamic change in user preferences and are crucial for predicting the next item. To tackle this issue, we propose a novel framework for mapping the sequential item texts to the sequential item IDs, named ST2SI. Specifically, we first introduce multi-query input and item linear projection (ILP) to model the conditional probability distribution of items. Then, we further propose ID alignment to address misalignment between item texts and item IDs by instruction tuning. Finally, we propose efficient ILP tuning to adapt flexibly to different scenarios, requiring only training a linear layer to achieve competitive performance. Extensive experiments on six real-world datasets show our approach outperforms the best baselines by 7.33% in NDCG@10, 4.65% in Recall@10, and 8.42% in MRR.

Cross-domain sequential recommendation (CSR) has garnered significant attention. Current federated frameworks for CSR leverage information across multiple domains but often rely on user alignment, which increases communication costs and privacy risks. In this work, we propose FedCSR, a novel federated cross-domain sequential recommendation framework that eliminates the need for user alignment between platforms. FedCSR fully utilizes cross-domain knowledge to address the key challenges related to data heterogeneity both inter- and intra-platform. To tackle the heterogeneity of data patterns between platforms, we introduce Model Contrastive Learning (MCL) to reduce the gap between local and global models. Additionally, we design Sequence Contrastive Learning (SCL) to address the heterogeneity of user preferences across different domains within a platform by employing tailored sequence augmentation techniques. Extensive experiments conducted on multiple real-world datasets demonstrate that FedCSR achieves superior performance compared to existing baseline methods.

Spike camera is a kind of neuromorphic camera with ultra-high temporal resolution, which can capture dynamic scenes by continuously firing spike signals. To capture color information, a color filter array (CFA) is employed on the sensor of the spike camera, resulting in Bayer-pattern spike streams. How to restore high-quality color images from the binary spike signals remains challenging. In this paper, we propose a motion-guided reconstruction method for spike cameras with CFA, utilizing color layout and estimated motion information. Specifically, we develop a joint motion estimation pipeline for the Bayer-pattern spike stream, exploiting the motion consistency of channels. We propose to estimate the missing pixels of each color channel according to temporally neighboring pixels of the corresponding color along the motion trajectory. As the spike signals are read out at discrete time points, there is quantization noise that impacts the image quality. Thus, we analyze the correlation of the noise in spatial and temporal domains and propose a self-supervised network utilizing a masked spike encoder to handle the noise. Experiments on real-world captured Bayer-pattern spike streams show that our method can restore color images with better visual quality, compared with state-of-the-art methods. The source codes are available at https://github.com/csycdong/SSL-CSC.

Indoor scene texture synthesis has garnered significant interest due to its important potential applications in virtual reality, digital media and creative arts. Existing diffusion-model-based researches either rely on per-view inpainting techniques, which are plagued by severe cross-view inconsistencies and conspicuous seams, or adopt optimization-based approaches that involve substantial computational overhead. In this work, we present RoomPainter, a frame-work that seamlessly integrates efficiency and consistency to achieve high-fidelity texturing of indoor scenes. The core of RoomPainter features a zero-shot technique that effectively adapts a 2D diffusion model for 3D-consistent texture synthesis, along with a two-stage generation strategy that ensures both global and local consistency. Specifically, we introduce Attention-Guided Multi-View Integrated Sampling (MVIS) combined with a neighbor-integrated attention mechanism for zero-shot texture map generation. Using the MVIS, we firstly generate texture map for the entire room to ensure global consistency, then adopt its variant, namely Attention-Guided Multi-View Integrated Repaint Sampling (MVRS) to repaint individual instances within the room, thereby further enhancing local consistency and addressing the occlusion problem. Experiments demonstrate that RoomPainter achieves superior performance for indoor scene texture synthesis in visual quality, global consistency and generation efficiency.

Vision-based Reinforcement Learning (VRL) attempts to establish associations between visual inputs and optimal actions through interactions with the environment. Given the high-dimensional and complex nature of visual data, it becomes essential to learn a policy based on high-quality state representation. To this end, existing VRL methods primarily rely on interaction-collected data, combined with selfsupervised auxiliary tasks. However, two key challenges remain: limited data samples and a lack of task-relevant semantic constraints. To tackle these challenges, we propose DGC, a method that Distills Guidance from Visual Language Models (VLMs) alongside self-supervised learning into a Compact VRL agent. Notably, we leverage the state representation capabilities of VLMs, rather than their decision-making abilities. Within DGC, a novel promptingreasoning pipeline is designed to convert historical observations and actions into usable supervision signals, enabling semantic understanding within the compact visual encoder. By leveraging these distilled semantic representations, the VRL agent achieves significant improvements in sample efficiency. Extensive experiments on the Carla benchmark demonstrate our state-of-the-art performance.

Integrating frames and events has become a widely accepted solution for various tasks in challenging scenarios. However, most multimodal methods directly convert events into image-like formats synchronized with frames and process each stream through separate two-branch backbones, making it difficult to fully exploit the spatiotemporal events while limiting inference frequency to the frame rate. To address these problems, we propose a novel asynchronous collaborative graph representation, namely ACGR, which is the first trial to explore a unified graph framework for asynchronously processing frames and events with high performance and low latency. Technically, we first construct uni-modal graphs for frames and events to preserve their spatiotemporal properties and sparsity. Then, an asynchronous collaborative alignment module is designed to align and fuse frames and events into a unified graph and the ACGR is generated through graph convolutional networks. Finally, we innovatively introduce domain adaptation to enable cross-modal interactions between frames and events by aligning their feature spaces. Experimental results show that our approach outperforms state-of-the-art methods in both object detection and depth estimation tasks, while significantly reducing computational latency and achieving real-time inference up to 200 Hz. Our code can be available at https://github.com/dianzl/ACGR.

Object detection in event streams has emerged as a cutting-edge research area, demonstrating superior performance in low-light conditions, scenarios with motion blur, and rapid movements. Current detectors leverage spiking neural networks, Transformers, or convolutional neural networks as their core architectures, each with its own set of limitations including restricted performance, high computational overhead, or limited local receptive fields. This paper introduces a novel MoE (Mixture of Experts) heat conduction-based object detection algorithm that strikingly balances accuracy and computational efficiency. Initially, we employ a stem network for event data embedding, followed by processing through our innovative MoE-HCO blocks. Each block integrates various expert modules to mimic heat conduction within event streams. Subsequently, an IoU-based query selection module is utilized for efficient token extraction, which is then channeled into a detection head for the final object detection process. Furthermore, we are pleased to introduce EvDET200K, a novel benchmark dataset for event-based object detection. Captured with a high-definition Prophesee EVK4-HD event camera, this dataset encompasses 10 distinct categories, 200,000 bounding boxes, and 10,054 samples, each spanning 2 to 5 seconds. We also provide comprehensive results from over 15 state-of-the-art detectors, offering a solid foundation for future research and comparison. The source code has been released on: https://github.com/Event-AHU/OpenEvDET

Recent improvements in visual synthesis have significantly enhanced the depiction of generated human photos, which are pivotal due to their wide applicability and demand. Nonetheless, the existing text-to-image or text-to-video models often generate low-quality human photos that might differ considerably from real-world body structures, referred to as "abnormal human bodies". Such abnormalities, typically deemed unacceptable, pose considerable challenges in the detection and repair of them within human photos. These challenges require precise abnormality recognition capabilities, which entail pinpointing both the location and the abnormality type. Intuitively, Visual Language Models (VLMs) that have obtained remarkable performance on various visual tasks are quite suitable for this task. However, their performance on abnormality detection in human photos is quite poor. Hence, it is quite important to highlight this task for the research community. In this paper, we first introduce a simple yet challenging task, i.e., Fine-grained Human-body Abnormality Detection (FHAD), and construct two high-quality datasets for evaluation. Then, we propose a meticulous framework, named HumanCalibrator, which identifies and repairs abnormalities in human body structures while preserving the other content. Experiments indicate that our HumanCalibrator achieves high accuracy in abnormality detection and accomplishes an increase in visual comparisons while preserving the other visual content.

Existing large language models (LLMs) evaluation methods typically focus on testing the performance on some closed-environment and domain-specific benchmarks with human annotations. In this paper, we explore a novel unsupervised evaluation direction, utilizing peer-review mechanisms to measure LLMs automatically without any human feedback. In this setting, both open-source and closed-source LLMs lie in the same environment, capable of answering unlabeled questions and evaluating each other, where each LLM’s response score is jointly determined by other anonymous ones. During this process, we found that those answers that are more recognized by other ``reviewers'' (models) usually come from LLMs with stronger abilities, while these models can also evaluate others' answers more accurately. We formalize it as a consistency assumption, i.e., the ability and score of the model usually have consistency. We exploit this to optimize each model's confidence, thereby re-ranking the LLMs to be closer to human rankings. We perform experiments on multiple datasets with standard rank-based metrics, validating the effectiveness of the proposed approach.

Generative recommendation has emerged as a promising paradigm aiming at directly generating the identifiers of the target candidates. Most existing methods attempt to leverage prior knowledge embedded in Pre-trained Language Models (PLMs) to improve the recommendation performance. However, they often fail to accommodate the differences between the general linguistic knowledge of PLMs and the specific needs of recommendation systems. Moreover, they rarely consider the complementary knowledge between the multimodal information of items, which represents the multi-faceted preferences of users. To facilitate efficient recommendation knowledge transfer, we propose a novel approach called Multimodal Quantitative Language for Generative Recommendation (MQL4GRec). Our key idea is to transform items from different domains and modalities into a unified language, which can serve as a bridge for transferring recommendation knowledge. Specifically, we first introduce quantitative translators to convert the text and image content of items from various domains into a new and concise language, known as quantitative language, with all items sharing the same vocabulary. Then, we design a series of quantitative language generation tasks to enrich quantitative language with semantic information and prior knowledge. Finally, we achieve the transfer of recommendation knowledge from different domains and modalities to the recommendation task through pre-training and fine-tuning. We evaluate the effectiveness of MQL4GRec through extensive experiments and comparisons with existing methods, achieving improvements over the baseline by 11.18\%, 14.82\%, and 7.95\% on the NDCG metric across three different datasets, respectively.

Incremental object detection is fundamentally challenged by catastrophic forgetting. A major factor contributing to this issue is background shift, where background categories in sequential tasks may overlap with either previously learned or future unseen classes. To address this, we propose a novel method called Class-Agnostic Shared Attribute Base (CASA) that encourages the model to learn category-agnostic attributes shared across incremental classes. Our approach leverages an LLM to generate candidate textual attributes, selects the most relevant ones based on the current training data, and records their importance in an assignment matrix. For subsequent tasks, the retained attributes are frozen, and new attributes are selected from the remaining candidates, ensuring both knowledge retention and adaptability. Extensive experiments on the COCO dataset demonstrate the state-of-the-art performance of our method.

Anomaly detection is essential for the safety and reliability of autonomous driving systems. Current methods often focus on detection accuracy but neglect response time, which is critical in time-sensitive driving scenarios. In this paper, we introduce real-time anomaly detection for autonomous driving, prioritizing both minimal response time and high accuracy. We propose a novel multimodal asynchronous hybrid network that combines event streams from event cameras with image data from RGB cameras. Our network utilizes the high temporal resolution of event cameras through an asynchronous Graph Neural Network and integrates it with spatial features extracted by a CNN from RGB images. This combination effectively captures both the temporal dynamics and spatial details of the driving environment, enabling swift and precise anomaly detection. Extensive experiments on benchmark datasets show that our approach outperforms existing methods in both accuracy and response time, achieving millisecond-level real-time performance.

Under the rapid advancement of large model-driven artificial intelligence, the surging energy consumption of AI training and inference tasks has created an urgent need for precise and comparable energy efficiency metrics to guide the design and deployment of green computing systems. While existing metrics such as PUE and Green500 metrics focus on infrastructure or traditional numerical computations, they cannot reflect the characteristics of AI workloads. Although applicationoriented metrics like J/response and J/token are designed for LLMs, they remain susceptible to biases induced by model scale and output strategies, lacking cross-model comparability. This paper proposes a novel AI energy efficiency metric, AUE, defined as the energy consumed per thousand tokens per billion activated parameters. By normalizing model size effects, AUE accurately reflects the energy efficiency of underlying computational resources. We theoretically justify the validity of AUE and conduct experiments on a server equipped with$4 \times$Ascend NPU 910C accelerators, evaluating dense Transformer and MoE architectures across both training and inference workloads. Experimental results demonstrate that traditional J/token metrics disproportionately favor smaller models, whereas AUE reveals true energy utilization efficiency. For instance, while Qwen3 0.6B shows superior J/token values compared to Qwen3 14B, the 14B model achieves a significantly better AUE of 12.88 J/(KToken GParam) versus 24.35 J/(KToken GParam) for the 0.6 B model, consistent with measured FLOPs where the 14B model outperforms its smaller counterpart. With advantages including simple measurement procedures and compatibility across platforms and model architectures, AUE provides a viable pathway toward establishing a normalized and standardized AI energy efficiency evaluation framework.

Novel view synthesis and 4D reconstruction techniques predominantly rely on RGB cameras, thereby inheriting inherent limitations such as the dependence on adequate lighting, susceptibility to motion blur, and a limited dynamic range. Event cameras, offering advantages of low power, high temporal resolution and high dynamic range, have brought a new perspective to addressing the scene reconstruction challenges in high-speed motion and low-light scenes. To this end, we propose E-4DGS, the first event-driven dynamic Gaussian Splatting approach, for novel view synthesis from multi-view event streams with fast-moving cameras. Specifically, we introduce an event-based initialization scheme to ensure stable training and propose event-adaptive slicing splatting for time-aware reconstruction. Additionally, we employ intensity importance pruning to eliminate floating artifacts and enhance 3D consistency, while incorporating an adaptive contrast threshold for more precise optimization. We design a synthetic multi-view camera setup with six moving event cameras surrounding the object in a 360-degree configuration and provide a benchmark multi-view event stream dataset that captures challenging motion scenarios. Our approach outperforms both event-only and event-RGB fusion baselines and paves the way for the exploration of multi-view event-based reconstruction as a novel approach for rapid scene capture.

Event-based object detection plays a crucial role in scenarios involving high-speed motion, extreme lighting conditions, and high-frequency detection. However, existing methods fail to address the challenges posed by small objects, including discriminative feature deficiency, the loss of critical information, and the inherent sparsity of event data. Moreover, the lack of benchmark datasets has significantly hindered progress in this field. To tackle these issues, we propose the Fully Deformable Detection Network (FDDNet), a lightweight framework that dynamically adapts to extract key features. First, we introduce a Long-Term Deformable Temporal Receptive Module (LDTR), which aligns critical features across consecutive event streams and leverages a State Space Model for long-range temporal modeling, enhancing the detection of high-speed small objects. Second, to address the sparsity of event data and the concentration of key features along object edges, we design a Sparse Feature Aggregation Block (SFAB) within the backbone and a coarse-to-fine deformable detection head, enabling hierarchical feature refinement from local to global, and improving the detection quality of sparse targets. Finally, to mitigate the lack of event-based small object datasets, we develop a high-quality, annotation-free data acquisition method and collect a real-world benchmark dataset for validation. Extensive experiments demonstrate that our approach achieves state-of-the-art (SOTA) performance on event-based small object detection tasks, with a mAP of 37.4% (+2.4%) on our benchmark and runs at 88 FPS, showcasing both accuracy and real-time capability. Our code and Supplement are available at https://github.com/Lqm26/ESOD.

Seeking high-quality representations with latent variable models (LVMs) to reveal the intrinsic correlation between neural activity and behavior or sensory stimuli has attracted much interest. In the study of the biological visual system, naturalistic visual stimuli are inherently high-dimensional and time-dependent, leading to intricate dynamics within visual neural activity. However, most work on LVMs has not explicitly considered neural temporal relationships. To cope with such conditions, we propose Time-Evolving Visual Dynamical System (TE-ViDS), a sequential LVM that decomposes neural activity into low-dimensional latent representations that evolve over time. To better align the model with the characteristics of visual neural activity, we split latent representations into two parts and apply contrastive learning to shape them. Extensive experiments on synthetic datasets and real neural datasets from the mouse visual cortex demonstrate that TE-ViDS achieves the best decoding performance on naturalistic scenes/movies, extracts interpretable latent trajectories that uncover clear underlying neural dynamics, and provides new insights into differences in visual information processing between subjects and between cortical regions. In summary, TE-ViDS is markedly competent in extracting stimulus-relevant embeddings from visual neural activity and contributes to the understanding of visual processing mechanisms. Our codes are available at https://github.com/Grasshlw/Time-Evolving-Visual-Dynamical-System.

While large language models (LLMs) with Chain-of-Thought (CoT) reasoning excel in mathematics and coding, their potential for systematic reasoning in chemistry, a domain demanding rigorous structural analysis for real-world tasks like drug design and reaction engineering, remains untapped. Current benchmarks focus on simple knowledge retrieval, neglecting step-by-step reasoning required for complex tasks such as molecular optimization and reaction prediction. To address this, we introduce ChemCoTBench, a reasoning framework that bridges molecular structure understanding with arithmetic-inspired operations, including addition, deletion, and substitution, to formalize chemical problem-solving into transparent, step-by-step workflows. By treating molecular transformations as modular "chemical operations", the framework enables slow-thinking reasoning, mirroring the logic of mathematical proofs while grounding solutions in real-world chemical constraints. We evaluate models on two high-impact tasks: Molecular Property Optimization and Chemical Reaction Prediction. These tasks mirror real-world challenges while providing structured evaluability. We further provide ChemCoTDataset, a pioneering 22,000-instance chemical reasoning dataset with expert-annotated chains of thought to facilitate LLM fine-tuning. By providing annotated trainable datasets, a reasoning taxonomy, and baseline evaluations, our work bridges the gap between abstract reasoning methods and practical chemical discovery, establishing a foundation for advancing LLMs as tools for AI-driven scientific innovation.

We introduce GS2E (Gaussian Splatting to Event Generation), a large-scale synthetic event dataset designed for high-fidelity event vision tasks, captured from real-world sparse multi-view RGB images. Existing event datasets are often synthesized from dense RGB videos, which typically suffer from limited viewpoint diversity and geometric inconsistency, or rely on expensive, hard-to-scale hardware setups. GS2E addresses these limitations by first reconstructing photorealistic static scenes using 3D Gaussian Splatting, followed by a novel, physically-informed event simulation pipeline. This pipeline integrates adaptive trajectory interpolation with physically-consistent event contrast threshold modeling. As a result, it generates temporally dense and geometrically consistent event streams under diverse motion and lighting conditions, while maintaining strong alignment with the underlying scene structure. Experimental results on event-based 3D reconstruction highlight GS2E’s superior generalization capabilities and its practical value as a benchmark for advancing event vision research.

Spiking Neural Networks (SNNs) are distinguished from Artificial Neural Networks (ANNs) for their complex neuronal dynamics and sparse binary activations (spikes) inspired by the biological neural system. Traditional neuron models use iterative step-by-step dynamics, resulting in serial computation and slow training speed of SNNs. Recently, parallelizable spiking neuron models have been proposed to fully utilize the massive parallel computing ability of graphics processing units to accelerate the training of SNNs. However, existing parallelizable spiking neuron models involve dense floating operations and can only achieve high long-term dependencies learning ability with a large order at the cost of huge computational and memory costs. To solve the dilemma of performance and costs, we propose the mul-free channel-wise Parallel Spiking Neuron, which is hardware-friendly and suitable for SNNs’ resource-restricted application scenarios. The proposed neuron imports the channel-wise convolution to enhance the learning ability, induces the sawtooth dilations to reduce the neuron order, and employs the bit-shift operation to avoid multiplications. The algorithm for the design and implementation of acceleration methods is discussed extensively. Our methods are validated in neuromorphic Spiking Heidelberg Digits voices, sequential CIFAR images, and neuromorphic DVS-Lip vision datasets, achieving superior performance over SOTA spiking neurons. Training speed results demonstrate the effectiveness of our acceleration methods, providing a practical reference for future research. Our code is available at Github.

Spike camera with high temporal resolution offers a new perspective on high-speed dynamic scene rendering. Most existing rendering methods rely on Neural Radiance Fields (NeRF) or 3D Gaussian Splatting (3DGS) for static scenes using a monocular spike camera. However, these methods struggle with dynamic motion, while a single camera suffers from limited spatial coverage, making it challenging to reconstruct fine details in high-speed scenes. To address these problems, we propose Spike4DGS, the first high-speed dynamic scene rendering framework with 4D Gaussian Splatting using spike camera arrays. Technically, we first build a multi-view spike camera array to validate our solution, then establish both synthetic and real-world multi-view spike-based reconstruction datasets. Then, we design a multi-view spike-based dense initialization module that obtains dense point clouds and camera poses from continuous spike streams. Finally, we propose a spike-pixel synergy constraint supervision to optimize Spike4DGS, incorporating both rendered image quality loss and dynamic spatiotemporal spike loss. The results show that our Spike4DGS outperforms state-of-the-art methods in terms of novel view rendering quality on both synthetic and real-world datasets. More details are available at https://github.com/Qinghongye/Spike4DGS.

Snapshot Mosaic Hyperspectral Cameras (SMHCs) are popular hyperspectral imaging devices for acquiring both color and motion details of scenes. However, the narrow-band spectral filters in SMHCs may negatively impact their motion perception ability, resulting in blurry SMHC frames. In this paper, we propose a hardware-software collaborative approach to address the blurring issue of SMHCs. Our approach involves integrating SMHCs with neuromorphic event cameras for efficient event-enhanced SMHC frame deblurring. To achieve spectral information recovery guided by event signals, we formulate a spectral-aware Event-based Double Integral (sEDI) model that links SMHC frames and events from a spectral perspective, providing principled model design insights. Then, we develop a Diffusion-guided Noise Awareness (DNA) training framework that utilizes diffusion models to learn noise-aware features and promote model robustness towards camera noise. Furthermore, we design an Event-enhanced Hyperspectral frame Deblurring Network (EvHDNet) based on sEDI, which is trained with DNA and features improved spatial-spectral learning and modality interaction for reliable SMHC frame deblurring. Experiments on both synthetic data and real data show that the proposed DNA + EvHDNet outperforms state-of-the-art methods on both spatial and spectral fidelity. The code and dataset will be made publicly available.

We focus on a very challenging task: imaging at nighttime dynamic scenes. Conventional RGB cameras struggle with the trade-off between long exposure for low-light imaging and short exposure for capturing dynamic scenes. Event cameras react to dynamic changes, with their high temporal resolution (microsecond) and dynamic range (120 dB), and thus offer a promising alternative. However, existing methods are mostly based on simulated datasets due to the lack of paired event-clean image data for nighttime conditions, where the domain gap leads to performance limitations in real-world scenarios. Moreover, most existing event reconstruction methods are tailored for daytime data, overlooking issues unique to low-light events at night, such as strong noise, temporal trailing, and spatial non-uniformity, resulting in unsatisfactory reconstruction results. To address these challenges, we construct the first real paired low-light event dataset (RLED) through a co-axial imaging system, comprising 80,400 spatially and temporally aligned image GTs and low-light events, which provides a unified training and evaluation dataset for existing methods. We further conduct a comprehensive analysis of the causes and characteristics of strong noise, temporal trailing, and spatial non-uniformity in nighttime events, and propose a nighttime event reconstruction network (NER-Net+). It includes a learnable event timestamps calibration module (LETC) to correct the temporal trailing events and a non-stationary spatio-temporal information enhancement module (NSIE) to suppress sensor noise and spatial non-uniformity. Extensive experiments demonstrate that the proposed method outperforms state-of-the-art methods in visual quality and generalization on real-world nighttime datasets.

This paper presents a Spatial Re-parameterization (SpRe) method for the N:M sparsity. SpRe stems from an observation regarding the restricted variety in spatial sparsity of convolution kernels presented in N:M sparsity compared with unstructured sparsity. Particularly, N:M sparsity exhibits a fixed sparsity rate within the spatial domains due to its distinctive pattern that mandates N non-zero components among M successive weights in the input channel dimension of convolution filters. On the contrary, we observe that conventional unstructured sparsity displays a substantial divergence in sparsity across the spatial domains, which we experimentally verify to be very crucial for its robust performance retention compared with N:M sparsity. Therefore, SpRe employs the spatial-sparsity distribution of unstructured sparsity by assigning an extra branch in conjunction with the original N:M branch at training time, which allows the N:M sparse network to sustain a similar distribution of spatial sparsity with unstructured sparsity. During inference, the extra branch can be further re-parameterized into the main N:M branch, without exerting any distortion on the sparse pattern or additional computation costs. SpRe has achieved a commendable feat by matching the performance of N:M sparsity methods with state-of-the-art unstructured sparsity methods across various benchmarks. Our project is available at https://github.com/zyxxmu/SpRE.

Depicting novel classes with language descriptions by observing few-shot samples is inherent in human-learning systems. This lifelong learning capability helps to distinguish new knowledge from old ones through the increase of open-world learning, namely Few-Shot Class-Incremental Learning (FSCIL). Existing works to solve this problem mainly rely on the careful tuning of visual encoders, which shows an evident trade-off between the base knowledge and incremental ones. Motivated by human learning systems, we propose a new Language-inspired Relation Transfer (LRT) paradigm to understand objects by joint visual clues and text depictions, composed of two major steps. We first transfer the pretrained text knowledge to the visual domains by proposing a graph relation transformation module and then fuse the visual and language embedding by a text-vision prototypical fusion module. Second, to mitigate the domain gap caused by visual finetuning, we propose context prompt learning for fast domain alignment and imagined contrastive learning to alleviate the insufficient text data during alignment. With collaborative learning of domain alignments and text-image transfer, our proposed LRT outperforms the state-of-the-art models by over 13% and 7% on the final session of miniImageNet and CIFAR-100 FSCIL benchmarks.

Optical flow has made great progress in clean scenes, while suffers degradation under adverse weather due to the violation of the brightness constancy and gradient continuity assumptions of optical flow. Typically, existing methods mainly adopt domain adaptation to transfer motion knowledge from clean to degraded domain through one-stage adaptation. However, this direct adaptation is ineffective, since there exists a large gap due to adverse weather and scene style between clean and real degraded domains. Moreover, even within the degraded domain itself, static weather (e.g., fog) and dynamic weather (e.g., rain) have different impacts on optical flow. To address above issues, we explore synthetic degraded domain as an intermediate bridge between clean and real degraded domains, and propose a cumulative homogeneous-heterogeneous adaptation framework for real adverse weather optical flow. Specifically, for clean-degraded transfer, our key insight is that static weather possesses the depth-association homogeneous feature which does not change the intrinsic motion of the scene, while dynamic weather additionally introduces the heterogeneous feature which results in a significant boundary discrepancy in warp errors between clean and degraded domains. For synthetic-real transfer, we figure out that cost volume correlation shares a similar statistical histogram between synthetic and real degraded domains, benefiting to holistically aligning the homogeneous correlation distribution for synthetic-real knowledge distillation. Under this unified framework, the proposed method can progressively and explicitly transfer knowledge from clean scenes to real adverse weather. In addition, we further collect a real adverse weather dataset with manually annotated optical flow labels and perform extensive experiments to verify the superiority of the proposed method.

Event cameras are novel bio-inspired sensors, where individual pixels operate independently and asynchronously, generating intensity changes as events. Leveraging the microsecond resolution (no motion blur) and high dynamic range (compatible with extreme light conditions) of events, there is considerable promise in directly segmenting objects from sparse and asynchronous event streams in various applications. However, different from the rich cues in video object segmentation, it is challenging to segment complete objects from the sparse event stream. In this paper, we present the first framework for continuous-time object segmentation from event stream. Given the object mask at the initial time, our task aims to segment the complete object at any subsequent time in event streams. Specifically, our framework consists of a Recurrent Temporal Embedding Extraction (RTEE) module based on a novel ResLSTM, a Cross-time Spatiotemporal Feature Modeling (CSFM) module which is a transformer architecture with long-term and short-term matching modules, and a segmentation head. The historical events and masks (reference sets) are recurrently fed into our framework along with current-time events. The temporal embedding is updated as new events are input, enabling our framework to continuously process the event stream. To train and test our model, we construct both real-world and simulated event-based object segmentation datasets, each comprising event streams, APS images, and object annotations. Extensive experiments on our datasets demonstrate the effectiveness of the proposed recurrent architecture.

Learned Image Compression (LIC) has attracted considerable attention due to their outstanding rate-distortion (R-D) performance and flexibility. However, the substantial computational cost poses challenges for practical deployment. The issue of feature redundancy in LIC is rarely addressed. Our findings indicate that many features within the LIC backbone network exhibit similarities. This paper introduces ShiftLIC, a novel and efficient LIC framework that employs parameter-free shift operations to replace large-kernel convolutions, significantly reducing the model’s computational burden and parameter count. Specifically, we propose the Spatial Shift Block (SSB), which combines shift operations with small-kernel convolutions to replace large-kernel. This approach maintains feature extraction efficiency while reducing both computational complexity and model size. To further enhance the representation capability in the channel dimension, we propose a channel attention module based on recursive feature fusion. This module enhances feature interaction while minimizing computational overhead. Additionally, we introduce an improved entropy model integrated with the SSB module, making the entropy estimation process more lightweight and thereby comprehensively reducing computational costs. Experimental results demonstrate that ShiftLIC outperforms leading compression methods, such as VVC Intra and GMM, in terms of computational cost, parameter count, and decoding latency. Additionally, ShiftLIC sets a new SOTA benchmark with a BD-rate gain per MACs/pixel of −102.6%, showcasing its potential for practical deployment in resource-constrained environments. The code is released athttps://github.com/baoyu2020/ShiftLIC.

Audio-visual zero-shot learning (ZSL) has been extensively researched for its capability to classify video data from unseen classes during training. Nevertheless, current methodologies often struggle with background scene biases and inadequate motion detail. This paper proposes a novel dual-stream Multi-Timescale Motion-Decoupled Spiking Transformer (MDST++), which decouples contextual semantic information and sparse dynamic motion information. The recurrent joint learning unit is proposed to extract contextual semantic information and capture joint knowledge across various modalities to understand the environment of actions. By converting RGB images to events, our method captures motion information more accurately and mitigates background scene biases. Moreover, we introduce a discrepancy analysis block to model audio motion information. To enhance the robustness of SNNs in extracting temporal and motion cues, we dynamically adjust the threshold of Leaky Integrate-and-Fire neurons based on global motion and contextual semantic information. Our experiments validate the effectiveness of MDST++, demonstrating their consistent superiority over state-of-the-art methods on mainstream benchmarks. Additionally, incorporating motion and multi-timescale information significantly improves HM and ZSL accuracy by 26.2% and 39.9%.

Bio-inspired vision sensors, which emulate the human retina by recording light intensity as binary spikes, have gained increasing interest in recent years. Among them, the spike camera is capable of perceiving fine textures by simulating a small retinal region called the fovea and producing high temporal resolution (20,000 Hz) spatiotemporal spike streams. To bridge the gap between binary spike streams and human vision in high-speed scenes, reconstructing intensity and optical flow from high temporal resolution spikes is particularly important. In this paper, we present a hybrid SNN-ANN network designed for simultaneous intensity and optical flow learning from spike streams. To adaptively extract spatial and temporal features from continuous spike streams, we propose a spiking neuron module with dense connections that efficiently processes both short-term and long-term spike data, while maintaining low power consumption characteristics. Subsequently, we introduce two decoders for optical flow and intensity estimation that complement each other. A temporal-aware warping module, based on flow features, is specifically designed to align the temporal features of the intensity decoder, thereby reducing motion artifacts. Concurrently, improved intensity features contribute to more accurate flow feature predictions, resulting in a mutually beneficial relationship within our network. To evaluate the effectiveness of our proposed network, we conduct experiments on both simulated and real spike datasets. Our network outperforms existing state-of-the-art spike-based reconstruction and optical flow estimation methods, demonstrating its potential for advancing the field of bio-inspired vision sensors. Our code is available athttps://github.com/LinZhu111/SLIO.

Current event stream-based pattern recognition models typically present the event stream as the point cloud, voxel, image, and the like, and formulate multiple deep neural networks to acquire their features. Although considerable results can be achieved in simple cases, however, the performance of the model might be restricted by monotonous modality expressions, sub-optimal fusion, and readout mechanisms. In this article, we put forward a novel dual-stream framework for event stream-based pattern recognition through differentiated fusion, which is called EFV++. It models two common event representations simultaneously, i.e., event images and event voxels. The spatial and three-dimensional stereo information can be separately learned by making use of Transformer and Graph Neural Network (GNN). We believe the features of each representation still contain both efficient and redundant features and a sub-optimal solution may be obtained if we directly fuse them without differentiation. Thus, we divide each feature into three levels and retain high-quality features, blend medium-quality features, and exchange low-quality features. The enhanced dual features will be provided to the fusion Transformer together with bottleneck features. In addition, we introduce a novel hybrid interaction readout mechanism to enhance the diversity of features as final representations. Comprehensive experiments validate that the framework we have proposed attains cutting-edge performance on a variety of extensively utilized event stream-based classification datasets. Particularly, we have realized a freshly pioneering performance on the Bullying10 k dataset, precisely 90.51%, and this outpaces the runner-up by$+2.21\%$.

The learned image compression (LIC) technique has surpassed the state-of-the-art traditional codecs (H.266/VVC) in case of rate-distortion (R-D) performance. Its real-time deployments are far advanced. In order to achieve more flexible deployments, an LIC technique should be flexible in adjusting its computational complexity and rate as demanded by a situation and its environment. In this paper, we propose a unified Rate-Distortion-Complexity (R-D-C) framework for LIC under channel energy concentration criteria. Specifically, we first introduce an Energy Asymptotic Nonlinear Transformation (EANT) designed to directly concentrate on the channel energy of latent representations, thus laying the groundwork for a scalable entropy coding. Next, leveraging this energy concentration characteristic, we propose a corresponding Heterogeneous Scalable Entropy Model (HSEM) for flexibly scaling bitstreams as needed. Finally, utilizing the proposed EANT, we construct a fine-grained scalable codec for formulating, in combination with HSEM, a comprehensive scalable R-D-C framework under the energy concentration criteria. The obtained experimental results demonstrate that the proposed method could enable seamless transitions between 13 different widths of sub-models within a single network, allowing for fine-grained control over the model bitrate, complexity, and hardware inference time. Additionally, the proposed method exhibits competitive R-D performance compared to many existing methods.

Existing datasets for RGB-DVS tracking are collected with DVS346 camera and their resolution ($346 \times 260$) is low for practical applications. Actually, only visible cameras are deployed in many practical systems, and the newly designed neuromorphic cameras may have different resolutions. The latest neuromorphic sensors can output high-definition event streams, but it is very difficult to achieve strict alignment between events and frames on both spatial and temporal views. Therefore, how to achieve accurate tracking with unaligned neuromorphic and visible sensors is a valuable but unresearched problem. In this work, we formally propose the task of object tracking using unaligned neuromorphic and visible cameras. We build the first unaligned frame-event dataset CRSOT collected with a specially built data acquisition system, which contains 1,030 high-definition RGB-Event video pairs, 304,974 video frames. In addition, we propose a novel unaligned object tracking framework that can realize robust tracking even using the loosely aligned RGB-Event data. This proposed method utilizes uncertainty perception techniques, which can effectively reduce the negative impact of noise (especially noise in event data) on tracking performance. Specifically, we extract the template and search regions of RGB and Event data and feed them into a unified ViT backbone for feature embedding. Next, we propose uncertainty perception modules to encode the RGB and Event features, respectively, then, we propose a modality uncertainty fusion module to aggregate the two modalities. These three branches are jointly optimized in the training phase. Extensive experiments demonstrate that our tracker can collaborate the dual modalities for high-performance tracking even without strictly temporal and spatial alignment.

With the help of special neuromorphic hardware, spiking neural networks (SNNs) are expected to realize artificial intelligence (AI) with less energy consumption. It provides a promising energy-efficient way for realistic control tasks by combining SNNs with deep reinforcement learning (DRL). In this article, we focus on the task where the agent needs to learn multidimensional deterministic policies to control, which is very common in real scenarios. Recently, the surrogate gradient method has been utilized for training multilayer SNNs, which allows SNNs to achieve comparable performance with the corresponding deep networks in this task. Most existing spike-based reinforcement learning (RL) methods take the firing rate as the output of SNNs, and convert it to represent continuous action space (i.e., the deterministic policy) through a fully connected (FC) layer. However, the decimal characteristic of the firing rate brings the floating-point matrix operations to the FC layer, making the whole SNN unable to deploy on the neuromorphic hardware directly. To develop a fully spiking actor network (SAN) without any floating-point matrix operations, we draw inspiration from the nonspiking interneurons found in insects and employ the membrane voltage of the nonspiking neurons to represent the action. Before the nonspiking neurons, multiple population neurons are introduced to decode different dimensions of actions. Since each population is used to decode a dimension of action, we argue that the neurons in each population should be connected in time domain and space domain. Hence, the intralayer connections are used in output populations to enhance the representation capacity. This mechanism exists extensively in animals and has been demonstrated effectively. Finally, we propose a fully SAN with intralayer connections (ILC-SAN). Extensive experimental results demonstrate that the proposed method outperforms the state-of-the-art performance on continuous control tasks from OpenAI gym. Moreover, we estimate the theoretical energy consumption when deploying ILC-SAN on neuromorphic chips to illustrate its high energy efficiency.

One important desideratum of lifelong learning aims to discover novel classes from unlabelled data in a continuous manner. The central challenge is twofold: discovering and learning novel classes while mitigating the issue of catastrophic forgetting of established knowledge. To this end, we introduce a new paradigm called Adaptive Discovering and Merging (ADM) to discover novel categories adaptively in the incremental stage and integrate novel knowledge into the model without affecting the original knowledge. To discover novel classes adaptively, we decouple representation learning and novel class discovery, and use Triple Comparison (TC) and Probability Regularization (PR) to constrain the probability discrepancy and diversity for adaptive category assignment. To merge the learned novel knowledge adaptively, we propose a hybrid structure with base and novel branches named Adaptive Model Merging (AMM), which reduces the interference of the novel branch on the old classes to preserve the previous knowledge, and merges the novel branch to the base model without performance loss and parameter growth. Extensive experiments on several datasets show that ADM significantly outperforms existing class-incremental Novel Class Discovery (class-iNCD) approaches. Moreover, our AMM also benefits the class-incremental Learning (class-IL) task by alleviating the catastrophic forgetting problem. The source code is included in the supplementary materials.

The main streams of human activity recognition (HAR) algorithms are developed based on RGB cameras which usually suffer from illumination, fast motion, privacy preservation, and large energy consumption. Meanwhile, the biologically inspired event cameras attracted great interest due to their unique features, such as high dynamic range, dense temporal but sparse spatial resolution, low latency, low power, etc. As it is a newly arising sensor, even there is no realistic large-scale dataset for HAR. Considering its great practical value, in this paper, we propose a large-scale benchmark dataset to bridge this gap, termed HARDVS, which contains 300 categories and more than 100K event sequences. We evaluate and report the performance of multiple popular HAR algorithms, which provide extensive baselines for future works to compare. More importantly, we propose a novel spatial-temporal feature learning and fusion framework, termed ESTF, for event stream based human activity recognition. It first projects the event streams into spatial and temporal embeddings using StemNet, then, encodes and fuses the dual-view representations using Transformer networks. Finally, the dual features are concatenated and fed into a classification head for activity prediction. Extensive experiments on multiple datasets fully validated the effectiveness of our model. Both the dataset and source code will be released at https://github.com/Event-AHU/HARDVS.

Generalized Category Discovery (GCD) aims to identify a mix of known and novel categories within unlabeled data sets, providing a more realistic setting for image recognition. Essentially, GCD needs to remember existing patterns thoroughly to recognize novel categories. Recent state-of-the-art method SimGCD transfers the knowledge from known-class data to the learning of novel classes through debiased learning. However, some patterns are catastrophically forgot during adaptation and thus lead to poor performance in novel categories classification. To address this issue, we propose a novel learning approach, LegoGCD, which is seamlessly integrated into previous methods to enhance the discrimination of novel classes while maintaining performance on previously encountered known classes. Specifically, we design two types of techniques termed as Local Entropy Regularization (LER) and Dual-views Kullback-Leibler divergence constraint (DKL). The LER optimizes the distribution of potential known class samples in unlabeled data, thus ensuring the preservation of knowledge related to known categories while learning novel classes. Meanwhile, DKL introduces Kullback-Leibler divergence to encourage the model to produce a similar prediction distribution of two view samples from the same image. In this way, it successfully avoids mismatched prediction and generates more reliable potential known class samples simultaneously. Extensive experiments validate that the proposed LegoGCD effectively addresses the known category forgetting issue across all datasets, e.g., delivering a 7.74% and 2.51% accuracy boost on known and novel classes in CUB, respectively. Our code is available at: https://github.com/Cliffia123/LegoGCD.

Visible and Infrared image Fusion (VIF) offers a comprehensive scene description by combining thermal infrared images with the rich textures from visible cameras. However, conventional VIF systems may capture over/under exposure or blurry images in extreme lighting and high dynamic motion scenarios, leading to degraded fusion results. To address these problems, we propose a novel Event-based Visible and Infrared Fusion (EVIF) system that employs a visible event camera as an alternative to traditional frame-based cameras for the VIF task. With extremely low latency and high dynamic range, event cameras can effectively address blurriness and are robust against diverse luminous ranges. To produce high-quality fused images, we develop a multitask collaborative framework that simultaneously performs event-based visible texture reconstruction, event-guided infrared image deblurring, and visible-infrared fusion. Rather than independently learning these tasks, our framework capitalizes on their synergy, leveraging cross-task event enhancement for efficient deblurring and bi-level min-max mutual information optimization to achieve higher fusion quality. Experiments on both synthetic and real data show that EVIF achieves remarkable performance in dealing with extreme lighting conditions and high-dynamic scenes, ensuring high-quality fused images across a broad range of practical scenarios.

Tracking with bio-inspired event cameras has garnered increasing interest in recent years. Existing works either utilize aligned RGB and event data for accurate tracking or directly learn an event-based tracker. The former incurs higher inference costs while the latter may be susceptible to the impact of noisy events or sparse spatial resolution. In this paper, we propose a novel hierarchical knowledge distillation framework that can fully utilize multimodal / multi-view information during training to facilitate knowledge transfer, enabling us to achieve high-speed and low-latency visual tracking during testing by using only event signals. Specifically, a teacher Transformer-based multimodal tracking framework is first trained by feeding the RGB frame and event stream simultaneously. Then, we design a new hierarchical knowledge distillation strategy which includes pairwise similarity, feature representation, and response maps-based knowledge distillation to guide the learning of the student Transformer network. In particular, since existing event-based tracking datasets are all low-resolution (346 × 260), we propose the first large-scale high-resolution (1280 × 720) dataset named EventVOT. It contains 1141 videos and covers a wide range of categories such as pedestrians, vehicles, UAVs, ping pong, etc. Ex-tensive experiments on both low-resolution (FE240hz, Vi-sEvent, COESOT), and our newly proposed high-resolution EventVOT dataset fully validated the effectiveness of our proposed method.

We explore visual reinforcement learning (RL) using two complementary visual modalities: frame-based RGB cam-era and event-based Dynamic Vision Sensor (DVS). Ex-isting multi-modality visual RL methods often encounter challenges in effectively extracting task-relevant information from multiple modalities while suppressing the in-creased noise, only using indirect reward signals instead of pixel-level supervision. To tackle this, we propose a Decomposed Multi-Modality Representation (DMR) framework for visual RL. It explicitly decomposes the inputs into three distinct components: combined task-relevant features (co-features), RGB-specific noise, and DVS-specific noise. The co-features represent the full information from both modalities that is relevant to the RL task; the two noise components, each constrained by a data reconstruction loss to avoid information leak, are contrasted with the co-features to maximize their difference. Extensive experiments demonstrate that, by explicitly separating the different types of information, our approach achieves substan-tially improved policy performance compared to state-of-the-art approaches.

Spiking Neural Networks (SNNs) have emerged as a promising third generation of neural networks, offering unique characteristics such as binary outputs, high sparsity, and biological plausibility. However, the lack of effective learning algorithms remains a challenge for SNNs. For instance, while converting artificial neural networks (ANNs) to SNNs circumvents the need for direct training of SNNs, it encounters issues related to conversion errors and high inference time delays. In order to reduce or even eliminate conversion errors while decreasing inference time-steps, we have introduced a novel type of neuron called Group Neurons (GNs). One GN is composed of multiple Integrate-and-Fire (IF) neurons as members, and its neural dynamics are meticulously designed. Based on GNs, we have optimized the traditional ANN-SNN conversion framework. Specifically, we replace the IF neurons in the SNNs obtained by the traditional conversion framework with GNs. The resulting SNNs, which utilize GNs, are capable of achieving accuracy levels comparable to ANNs even within extremely short inference time-steps. The experiments on CIFAR10, CIFAR100, and ImageNet datasets demonstrate the superiority of the proposed methods in terms of both inference accuracy and latency. Code is available at https://github.com/Lyu6PosHao/ANN2SNN_GN.

Neuromorphic computing, which exploits Spiking Neural Networks (SNNs) on neuromorphic chips, is a promising energy-efficient alternative to traditional AI. CNN-based SNNs are the current mainstream of neuromorphic computing. By contrast, no neuromorphic chips are designed especially for Transformer-based SNNs, which have just emerged, and their performance is only on par with CNN-based SNNs, offering no distinct advantage. In this work, we propose a general Transformer-based SNN architecture, termed as ``Meta-SpikeFormer", whose goals are: (1) *Lower-power*, supports the spike-driven paradigm that there is only sparse addition in the network; (2) *Versatility*, handles various vision tasks; (3) *High-performance*, shows overwhelming performance advantages over CNN-based SNNs; (4) *Meta-architecture*, provides inspiration for future next-generation Transformer-based neuromorphic chip designs. Specifically, we extend the Spike-driven Transformer in \citet{yao2023spike} into a meta architecture, and explore the impact of structure, spike-driven self-attention, and skip connection on its performance. On ImageNet-1K, Meta-SpikeFormer achieves 80.0\% top-1 accuracy (55M), surpassing the current state-of-the-art (SOTA) SNN baselines (66M) by 3.7\%. This is the first direct training SNN backbone that can simultaneously supports classification, detection, and segmentation, obtaining SOTA results in SNNs. Finally, we discuss the inspiration of the meta SNN architecture for neuromorphic chip design.

Deep neural networks (DNNs) are widely used models for investigating biological visual representations. However, existing DNNs are mostly designed to analyze neural responses to static images, relying on feedforward structures and lacking physiological neuronal mechanisms. There is limited insight into how the visual cortex represents natural movie stimuli that contain context-rich information. To address these problems, this work proposes the long-range feedback spiking network (LoRaFB-SNet), which mimics top-down connections between cortical regions and incorporates spike information processing mechanisms inherent to biological neurons. Taking into account the temporal dependence of representations under movie stimuli, we present Time-Series Representational Similarity Analysis (TSRSA) to measure the similarity between model representations and visual cortical representations of mice. LoRaFB-SNet exhibits the highest level of representational similarity, outperforming other well-known and leading alternatives across various experimental paradigms, especially when representing long movie stimuli. We further conduct experiments to quantify how temporal structures (dynamic information) and static textures (static information) of the movie stimuli influence representational similarity, suggesting that our model benefits from long-range feedback to encode context-dependent representations just like the brain. Altogether, LoRaFB-SNet is highly competent in capturing both dynamic and static representations of the mouse visual cortex and contributes to the understanding of movie processing mechanisms of the visual system. Our codes are available at https://github.com/Grasshlw/SNN-Neural-Similarity-Movie.

Accurate environment dynamics modeling is crucial for obtaining effective state representations in visual reinforcement learning (RL) applications. However, when facing multiple input modalities, existing dynamics modeling methods (e.g., DeepMDP) usually stumble in addressing the complex and volatile relationship between different modalities. In this paper, we study the problem of efficient dynamics modeling for multi-modal visual RL. We find that under the existence of modality heterogeneity, modality-correlated and distinct features are equally important but play different roles in reflecting the evolution of environmental dynamics. Motivated by this fact, we propose Dissected Dynamics Modeling (DDM), a novel multi-modal dynamics modeling method for visual RL. Unlike existing methods, DDM explicitly distinguishes consistent and inconsistent information across modalities and treats them separately with a divide-and-conquer strategy. This is done by dispatching the features carrying different information into distinct dynamics modeling pathways, which naturally form a series of implicit regularizations along the learning trajectories. In addition, a reward predictive function is further introduced to filter task-irrelevant information in both modality-consistent and inconsistent features, ensuring information integrity while avoiding potential distractions. Extensive experiments show that DDM consistently achieves competitive performance in challenging multi-modal visual environments.

Spiking Transformers, which integrate Spiking Neural Networks (SNNs) with Transformer architectures, have attracted significant attention due to their potential for low energy consumption and high performance. However, there remains a substantial gap in performance between SNNs and Artificial Neural Networks (ANNs). To narrow this gap, we have developed QKFormer, a direct training spiking transformer with the following features: i) _Linear complexity and high energy efficiency_, the novel spike-form Q-K attention module efficiently models the token or channel attention through binary vectors and enables the construction of larger models. ii) _Multi-scale spiking representation_, achieved by a hierarchical structure with the different numbers of tokens across blocks. iii) _Spiking Patch Embedding with Deformed Shortcut (SPEDS)_, enhances spiking information transmission and integration, thus improving overall performance. It is shown that QKFormer achieves significantly superior performance over existing state-of-the-art SNN models on various mainstream datasets. Notably, with comparable size to Spikformer (66.34 M, 74.81\%), QKFormer (64.96 M) achieves a groundbreaking top-1 accuracy of **85.65\%** on ImageNet-1k, substantially outperforming Spikformer by **10.84\%**. To our best knowledge, this is the first time that directly training SNNs have exceeded 85\% accuracy on ImageNet-1K.

Mix-precision computation is crucial for artificial intelligence and scientific computing applications. However, as novel chips with innovative architectures emerge, harnessing their computational capabilities presents significant challenges. While existing algorithms for the HPL-MxP LU factorization excel on homogeneous systems, they often encounter difficulties on specialized heterogeneous architectures. This deficiency arises from inadequate optimization for computation, memory access, and communication, hindering effective mixed-precision acceleration. This work introduces an algorithm-hardware co-optimization approach for LU factorization on specialized NPUs and CPUs, leveraging their unique architectures. A novel multi-iteration fusion method for general matrix multiplication is proposed, strategically designed to maximize on-chip L1 buffer utilization, effectively overcoming the notorious “memory wall”. Additionally, a multi-stage, multi-level heterogeneous pipeline for LU factorization in an accelerator-CPU cloud environment is presented, where compute-intensive matrix multiplications are offloaded to NPUs while CPUs handle the remaining tasks. The co-optimization approach fosters deep collaboration between CPUs and accelerators, thereby unlocking enhanced performance.

Most existing learning-based deraining methods are supervisedly trained on synthetic rainy-clean pairs. The domain gap between the synthetic and real rain makes them less generalized to complex real rainy scenes. Moreover, the existing methods mainly utilize the property of the image or rain layers independently, while few of them have considered their mutually exclusive relationship. To solve above dilemma, we explore the intrinsic intra-similarity within each layer and inter-exclusiveness between two layers and propose an unsupervised non-local contrastive learning (NLCL) deraining method. The non-local self-similarity image patches as the positives are tightly pulled together and rain patches as the negatives are remarkably pushed away, and vice versa. On one hand, the intrinsic self-similarity knowledge within positive/negative samples of each layer benefits us to discover more compact representation; on the other hand, the mutually exclusive property between the two layers enriches the discriminative decomposition. Thus, the internal self-similarity within each layer (similarity) and the external exclusive relationship of the two layers (dissimilarity) serving as a generic image prior jointly facilitate us to unsupervisedly differentiate the rain from clean image. We further discover that the intrinsic dimension of the non-local image patches is generally higher than that of the rain patches. This insight motivates us to design an asymmetric contrastive loss that precisely models the compactness discrepancy of the two layers, thereby improving the discriminative decomposition. In addition, recognizing the limited quality of existing real rain datasets, which are often small-scale or obtained from the internet, we collect a large-scale real dataset under various rainy weathers that contains high-resolution rainy images. Extensive experiments conducted on different real rainy datasets demonstrate that the proposed method obtains state-of-the-art performance in real deraining.

PSNR-oriented models are a critical class of super-resolution models with applications across various fields. However, these models tend to generate over-smoothed images, a problem that has been analyzed previously from the perspectives of models or loss functions, but without taking into account the impact of data properties. In this paper, we present a novel phenomenon that we term the center-oriented optimization (COO) problem, where a model's output converges towards the center point of similar high-resolution images, rather than towards the ground truth. We demonstrate that the strength of this problem is related to the uncertainty of data, which we quantify using entropy. We prove that as the entropy of high-resolution images increases, their center point will move further away from the clean image distribution, and the model will generate over-smoothed images. Implicitly optimizing the COO problem, perceptual-driven approaches such as perceptual loss, model structure optimization, or GAN-based methods can be viewed. We propose an explicit solution to the COO problem, called Detail Enhanced Contrastive Loss (DECLoss). DECLoss utilizes the clustering property of contrastive learning to directly reduce the variance of the potential high-resolution distribution and thereby decrease the entropy. We evaluate DECLoss on multiple super-resolution benchmarks and demonstrate that it improves the perceptual quality of PSNR-oriented models. Moreover, when applied to GAN-based methods, such as RaGAN, DECLoss helps to achieve state-of-the-art performance, such as 0.093 LPIPS with 24.51 PSNR on 4× downsampled Urban100, validating the effectiveness and generalization of our approach.

Transformers have shown remarkable performance, however, their architecture design is a time-consuming process that demands expertise and trial-and-error. Thus, it is worthwhile to investigate efficient methods for automatically searching high-performance Transformers via Transformer Architecture Search (TAS). In order to improve the search efficiency, training-free proxy based methods have been widely adopted in Neural Architecture Search (NAS). Whereas, these proxies have been found to be inadequate in generalizing well to Transformer search spaces, as confirmed by several studies and our own experiments. This paper presents an effective scheme for TAS called TRansformer Architecture search with ZerO-cost pRoxy guided evolution (T-Razor) that achieves exceptional efficiency. First, through theoretical analysis, we discover that the synaptic diversity of multi-head self-attention (MSA) and the saliency of multi-layer perceptron (MLP) are correlated with the performance of corresponding Transformers. The properties of synaptic diversity and synaptic saliency motivate us to introduce the ranks of synaptic diversity and saliency that denoted as DSS++ for evaluating and ranking Transformers. DSS++ incorporates correlation information among sampled Transformers to provide unified scores for both synaptic diversity and synaptic saliency. We then propose a block-wise evolution search guided by DSS++ to find optimal Transformers. DSS++ determines the positions for mutation and crossover, enhancing the exploration ability. Experimental results demonstrate that our T-Razor performs competitively against the state-of-the-art manually or automatically designed Transformer architectures across four popular Transformer search spaces. Significantly, T-Razor improves the searching efficiency across different Transformer search spaces, e.g., reducing required GPU days from more than 24 to less than 0.4 and outperforming existing zero-cost approaches. We also apply T-Razor to the BERT search space and find that the searched Transformers achieve competitive GLUE results on several Neural Language Processing (NLP) datasets. This work provides insights into training-free TAS, revealing the usefulness of evaluating Transformers based on the properties of their different blocks.

Brain-inspired computing (BIC) is an emerging research field that aims to build fundamental theories, models, hardware architectures, and application systems toward more general artificial intelligence (AI) by learning from the information processing mechanisms or structures/functions of biological nervous systems. It is regarded as one of the most promising research directions for future intelligent computing in the post-Moore era. In the past few years, various new schemes in this field have sprung up to explore more general AI. These works are quite divergent in the aspects of modeling/algorithm, software tool, hardware platform, and benchmark data since BIC is an interdisciplinary field that consists of many different domains, including computational neuroscience, AI, computer science, statistical physics, material science, and microelectronics. This situation greatly impedes researchers from obtaining a clear picture and getting started in the right way. Hence, there is an urgent requirement to do a comprehensive survey in this field to help correctly recognize and analyze such bewildering methodologies. What are the key issues to enhance the development of BIC? What roles do the current mainstream technologies play in the general framework of BIC? Which techniques are truly useful in real-world applications? These questions largely remain open. To address the above issues, in this survey, we first clarify the biggest challenge of BIC: how can AI models benefit from the recent advancements in computational neuroscience? With this challenge in mind, we will focus on discussing the concept of BIC and summarize four components of BIC infrastructure development: 1) modeling/algorithm; 2) hardware platform; 3) software tool; and 4) benchmark data. For each component, we will summarize its recent progress, main challenges to resolve, and future trends. Based on these studies, we present a general framework for the real-world applications of BIC systems, which is promising to benefit both AI and brain science. Finally, we claim that it is extremely important to build a research ecology to promote prosperity continuously in this field.

Presents corrections to the paper, (Corrections to “Brain-Inspired Computing: A Systematic Survey and Future Trends”).

This paper proposes a unified and efficient entropy coding method for learned image compression (LIC) from the perspective of traditional signal processing. First, the consistency of structures and optimization objectives are used to interpret the existing split-coded-then-merge entropy coding strategies in LIC as a particular filter banks framework, with feature separation and feature aggregation representing the analysis filter bank and synthesis filter bank, respectively. Thus, we borrow the design from the multirate filter banks and proposed Multirate Progressive Entropy Model (MPEM) to enhance the rate-distortion performance and decoding speed. In particular, we create an analysis filter bank that divides compact features into a few nonuniform subsets based on various spatial and channel sampling rates. Then multi-scale detail and mean coefficients within the current subset are used as prior representations to help generate the prediction parameters of the next subset, and the carefully designed synthetic filter bank performs a near-perfect reconstruction of the features. In addition, we propose a Multi-level Edge Attention Moudal (MEAM) to increase the edge and texture information’s contribution and reduce the high-frequency information loss brought on by MPEM’s inherent multi-rate spatial sampling, which leverages the edge operator and structural reparameterization principles. The results of the experiments show that, in comparison to the effective LIC methods and traditional code, the proposed MPEM can decode data at a cutting-edge speed while also offering comparable rate-distortion performance.

Event cameras, offering high temporal resolutions and high dynamic ranges, have brought a new perspective to address common challenges in monocular depth estimation (e.g., motion blur and low light). However, existing CNN-based methods insufficiently exploit global spatial information from asynchronous events, while RNN-based methods show a limited capacity for effective temporal cues utilization for event-based monocular depth estimation. To this end, we propose a event-based monocular depth estimator with recurrent transformers, namely EReFormer. Technically, we first design a transformer-based encoder-decoder that utilizes multi-scale features to model global spatial information from events. Then, we propose a Gate Recurrent Vision Transformer (GRViT), introducing a recursive mechanism into transformers, to leverage rich temporal cues from events. Finally, we present a Cross Attention-guided Skip Connection (CASC), performing cross attention to fuse multi-scale features, to improve global spatial modeling capabilities. The experimental results show that our EReFormer outperforms state-of-the-art methods by a margin on both synthetic and real-world datasets. Our open-source code is available at https://github.com/liuxu0303/EReFormer.

Different from visible cameras which record intensity images frame by frame, the biologically inspired event camera produces a stream of asynchronous and sparse events with much lower latency. In practice, visible cameras can better perceive texture details and slow motion, while event cameras can be free from motion blurs and have a larger dynamic range which enables them to work well under fast motion and low illumination (LI). Therefore, the two sensors can cooperate with each other to achieve more reliable object tracking. In this work, we propose a large-scale Visible-Event benchmark (termed VisEvent) due to the lack of a realistic and scaled dataset for this task. Our dataset consists of 820 video pairs captured under LI, high speed, and background clutter scenarios, and it is divided into a training and a testing subset, each of which contains 500 and 320 videos, respectively. Based on VisEvent, we transform the event flows into event images and construct more than 30 baseline methods by extending current single-modality trackers into dual-modality versions. More importantly, we further build a simple but effective tracking algorithm by proposing a cross-modality transformer, to achieve more effective feature fusion between visible and event data. Extensive experiments on the proposed VisEvent dataset, FE108, COESOT, and two simulated datasets (i.e., OTB-DVS and VOT-DVS), validated the effectiveness of our model. The dataset and source code have been released on: https://github.com/wangxiao5791509/VisEvent_SOT_Benchmark.

Model inversion attacks can reconstruct the training samples of victim deep learning models. The existing efforts heavily rely on auxiliary information of the target samples (prior target information) to achieve their adversarial goals. However, prior target information is hard to obtain in practice. In this paper, we explore the effect of class information in model inversion attacks to reduce the reliance of prior target information. Our contributions on class information exploitation are two-fold. Firstly, we propose a supervised inversion model, Supervised Model Inversion (SMI). The proposed inversion model learns pixel-level features and data-to-class features from the rounded-outputs of the victim model and labeled auxiliary dataset. Secondly, we leverage victim model's rounded-outputs to guide the optimization of reconstructing inversion samples after trained inversion model. Our experimental results show that inversion samples reconstructed by SMI are more visually plausible with more details, comparing to the three representative model inversion attacks. We further perform an extensive study on various auxiliary dataset settings. It is found that the class combination in the auxiliary dataset rather than the number of classes that determines the quality of inversion samples. The ground-truth labels can improve the qualities of inversion samples but not essential to inversion attacks.

With the benefit of deep learning techniques, recent researches have made significant progress in image compression artifacts reduction. Despite their improved performances, prevailing methods only focus on learning a mapping from the compressed image to the original one but ignore the intrinsic attributes of the given compressed images, which greatly harms the performance of downstream parsing tasks. Different from these methods, we propose to decouple the intrinsic attributes into two complementary features for artifacts reduction, i.e., the compression-insensitive features to regularize the high-level semantic representations during training and the compression-sensitive features to be aware of the compression degree. To achieve this, we first employ adversarial training to regularize the compressed and original encoded features for retaining high-level semantics, and we then develop the compression quality-aware feature encoder for compression-sensitive features. Based on these dual complementary features, we propose a Dual Awareness Guidance Network (DAGN) to utilize these awareness features as transformation guidance during the decoding phase. In our proposed DAGN, we develop a cross-feature fusion module to maintain the consistency of compression-insensitive features by fusing compression-insensitive features into the artifacts reduction baseline. Our method achieves an average 2.06 dB PSNR gains on BSD500, outperforming state-of-the-art methods, and only requires 29.7 ms to process one image on BSD500. Besides, the experimental results on LIVE1 and LIU4K also demonstrate the efficiency, effectiveness, and superiority of the proposed method in terms of quantitative metrics, visual quality, and downstream machine vision tasks.

To manipulate large-scale data, anchor-based multi-view clustering methods have grown in popularity owing to their linear complexity in terms of the number of samples. However, these existing approaches pay less attention to two aspects. 1) They target at learning a shared affinity matrix by using the local information from every single view, yet ignoring the global information from all views, which may weaken the ability to capture complementary information. 2) They do not consider the removal of feature redundancy, which may affect the ability to depict the real sample relationships. To this end, we propose a novel fast multi-view clustering method via pick-and-place transform learning named PPTL, which could capture insightful global features to characterize the sample relationships quickly. Specifically, PPTL first concatenates all the views along the feature direction to produce a global matrix. Considering the redundancy of the global matrix, we design a pick-and-place transform with ℓ2,p-norm regularization to abandon the poor features and consequently construct a compact global representation matrix. Thus, by conducting anchor-based subspace clustering on the compact global representation matrix, PPTL can learn a consensus skinny affinity matrix with a discriminative clustering structure. Numerous experiments performed on small-scale to large-scale datasets demonstrate that our method is not only faster but also achieves superior clustering performance over state-of-the-art methods across a majority of the datasets.

Nowadays, Connected Automated Vehicles (CAVs) have emerged as powerful infrastructures for the next-generation Intelligent Transportation System (ITS) as the rapid technological advancements of communication networks and vehicular intelligence. While prospective platoon-based techniques in CAVs, the heterogeneous traffic condition poses a challenge for platoon control in the self-organized traffic bottleneck, thus making an urgent need for a practical sustainable transportation architecture. To address this problem, we propose a software defined architecture that leverages multi-agent techniques to mobile-edge computing networks for multi-vehicle adaptive platoon, which is called SD-M3ASP. The architecture supports centralized and decentralized management of vehicular edge communication resources between mobile vehicles and edge devices, and underpins sustainable vehicular platooning capabilities. Then, we propose cluster-based kinematic models by grouping vehicles into multi-vehicle clusters (MVCs) to facilitate efficient platoon control with collision avoidance. Furthermore, we propose three-stage platoon control algorithms to adaptively balance the size of MVCs and form stable platoons in heterogeneous traffic flows. The intra-platoon and inter-platoon convergence are analyzed by using the Routh stability criterion and Lyapunov technique. A CAV simulation software is developed for demonstration purposes which is available online athttps://qgailab.com/cav-sim. Extensive numerical simulation results have shown the superiority of the proposed method, which can greatly eliminate the self-organized congestion caused by heterogeneous traffic flow.

With the rapid development of wireless communication and localization technologies, the easier collection of trajectory data can bring potential data-driven value. Recently, there has been an increasing interest in how to publish trajectory dataset without revealing personal information. However, since the large-scale and real-world sequential trajectory dataset presents a heterogeneous regional distribution, the existing study ignores the relationship between privacy budget allocation and spatial characteristics, resulting in unreasonable continuity and mapping distortion, and thus lowering the utility of the synthetic dataset. To address this problem, we propose a probability distribution model named Adaptive grid-based Weighted Differential Privacy (AWDP). First, trajectories are adaptively discretized into the multi-resolution grid structures to make trajectories more uniformly distributed and less disturbed by the noise. Second, we allocate different weighted budgets for different grids according to density-based regional characteristics. Third, a spatio-temporal continuity maintenance method is designed to solve unrealistic direction- and density-based continuity deviations of synthetic trajectories. An application system is developed for demonstration purposes which is available online athttp://qgailab.com/awdp/. The extensive experiments on three datasets demonstrate that AWDP performs significantly better than the state-of-the-art model in preserving the density distribution of the original trajectories with differential privacy guarantee and high utility.

Data-dependent hashing methods aim to learn hash functions from the pairwise or triplet relationships among the data, which often lead to low efficiency and low collision rate by only capturing the local distribution of the data. To solve the limitation, we propose central similarity, in which the hash codes of similar data pairs are encouraged to approach a common center and those of dissimilar pairs to converge to different centers. As a new global similarity metric, central similarity can improve the efficiency and retrieval accuracy of hash learning. By introducing a new concept, hash centers, we principally formulate the computation of the proposed central similarity metric, in which the hash centers refer to a set of points scattered in the Hamming space with a sufficient mutual distance between each other. To construct well-separated hash centers, we provide two efficient methods: 1) leveraging the Hadamard matrix and Bernoulli distributions to generate data-independent hash centers and 2) learning data-dependent hash centers from data representations. Based on the proposed similarity metric and hash centers, we propose central similarity quantization (CSQ) that optimizes the central similarity between data points with respect to their hash centers instead of optimizing the local similarity to generate a high-quality deep hash function. We also further improve the CSQ with data-dependent hash centers, dubbed as CSQ with learnable center (CSQLC). The proposed CSQ and CSQLC are generic and applicable to image and video hashing scenarios. We conduct extensive experiments on large-scale image and video retrieval tasks, and the proposed CSQ yields noticeably boosted retrieval performance, i.e., 3%-20% in mean average precision (mAP) over the previous state-of-the-art methods, which also demonstrates that our methods can generate cohesive hash codes for similar data pairs and dispersed hash codes for dissimilar pairs.

Deep artificial neural networks (ANNs) play a major role in modeling the visual pathways of primate and rodent. However, they highly simplify the computational properties of neurons compared to their biological counterparts. Instead, Spiking Neural Networks (SNNs) are more biologically plausible models since spiking neurons encode information with time sequences of spikes, just like biological neurons do. However, there is a lack of studies on visual pathways with deep SNNs models. In this study, we model the visual cortex with deep SNNs for the first time, and also with a wide range of state-of-the-art deep CNNs and ViTs for comparison. Using three similarity metrics, we conduct neural representation similarity experiments on three neural datasets collected from two species under three types of stimuli. Based on extensive similarity analyses, we further investigate the functional hierarchy and mechanisms across species. Almost all similarity scores of SNNs are higher than their counterparts of CNNs with an average of 6.6%. Depths of the layers with the highest similarity scores exhibit little differences across mouse cortical regions, but vary significantly across macaque regions, suggesting that the visual processing structure of mice is more regionally homogeneous than that of macaques. Besides, the multi-branch structures observed in some top mouse brain-like neural networks provide computational evidence of parallel processing streams in mice, and the different performance in fitting macaque neural representations under different stimuli exhibits the functional specialization of information processing in macaques. Taken together, our study demonstrates that SNNs could serve as promising candidates to better model and explain the functional hierarchy and mechanisms of the visual system.

Recent advances in 3D point cloud analysis bring a diverse set of network architectures to the field. However, the lack of a unified framework to interpret those networks makes any systematic comparison, contrast, or analysis challenging, and practically limits healthy development of the field. In this paper, we take the initiative to explore and propose a unified framework called PointMeta, to which the popular 3D point cloud analysis approaches could fit. This brings three benefits. First, it allows us to compare different approaches in a fair manner, and use quick experiments to verify any empirical observations or assumptions summarized from the comparison. Second, the big picture brought by PointMeta enables us to think across different components, and revisit common beliefs and key design decisions made by the popular approaches. Third, based on the learnings from the previous two analyses, by doing simple tweaks on the existing approaches, we are able to derive a basic building block, termed PointMetaBase. It shows very strong performance in efficiency and effectiveness through extensive experiments on challenging benchmarks, and thus verifies the necessity and benefits of high-level interpretation, contrast, and comparison like PointMeta. In particular, PointMetaBase surpasses the previous state-of-the-art method by 0.7%/1.4/%2.1% mIoU with only 2%/11%/13% of the computation cost on the S3DIS datasets. The code and models are available at https://github.com/linhaojia13/PointMetaBase.

Few-shot class-incremental learning (FSCIL) aims at learning to classify new classes continually from limited samples without forgetting the old classes. The mainstream framework tackling FSCIL is first to adopt the cross-entropy (CE) loss for training at the base session, then freeze the feature extractor to adapt to new classes. However, in this work, we find that the CE loss is not ideal for the base session training as it suffers poor class separation in terms of representations, which further degrades generalization to novel classes. One tempting method to mitigate this problem is to apply an additional naïve supervised contrastive learning (SCL) in the base session. Unfortunately, we find that although SCL can create a slightly better representation separation among different base classes, it still struggles to separate base classes and new classes. Inspired by the observations made, we propose Semantic-Aware Virtual Contrastive model (SAVC), a novel method that facilitates separation between new classes and base classes by introducing virtual classes to SCL. These virtual classes, which are generated via pre-defined transformations, not only act as placeholders for unseen classes in the representation space, but also provide diverse semantic information. By learning to recognize and contrast in the fantasy space fostered by virtual classes, our SAVC significantly boosts base class separation and novel class generalization, achieving new state-of-the-art performance on the three widely-used FSCIL benchmark datasets. Code is available at: https://github.com/zysong0113/SAVC.

Efficient motion and appearance modeling are critical for vision-based Reinforcement Learning (RL). However, existing methods struggle to reconcile motion and appearance information within the state representations learned from a single observation encoder. To address the problem, we present Synergizing Interactive Motion-appearance Understanding (Simoun), a unified framework for vision-based RL Given consecutive observation frames, Simoun deliberately and interactively learns both motion and appearance features through a dual-path network architecture. The learning process collaborates with a structural interactive module, which explores the latent motion-appearance structures from the two network paths to leverage their complementarity. To promote sample efficiency, we further design a consistency-guided curiosity module to encourage the exploration of under-learned observations. During training, the curiosity module provides intrinsic rewards according to the consistency of environmental temporal dynamics, which are deduced from both motion and appearance network paths. Experiments conducted on Deep-Mind control suite and CARLA automatic driving benchmarks demonstrate the effectiveness of Simoun, where it performs favorably against state-of-the-art methods.

Vision-based reinforcement learning (RL) depends on discriminative representation encoders to abstract the observation states. Despite the great success of increasing CNN parameters for many supervised computer vision tasks, reinforcement learning with temporal-difference (TD) losses cannot benefit from it in most complex environments. In this paper, we analyze that the training instability arises from the oscillating self-overfitting of the heavy-optimizable encoder. We argue that serious oscillation will occur to the parameters when enforced to fit the sensitive TD targets, causing uncertain drifting of the latent state space and thus transmitting these perturbations to the policy learning. To alleviate this phenomenon, we propose a novel asymmetric interactive cooperation approach with the interaction between a heavy-optimizable encoder and a supportive light-optimizable encoder, in which both their advantages are integrated including the highly discriminative capability as well as the training stability. We also present a greedy bootstrapping optimization to isolate the visual perturbations from policy learning, where representation and policy are trained sufficiently by turns. Finally, we demonstrate the effectiveness of our method in utilizing larger visual models by first-person highway driving task CARLA and Vizdoom environments.

Decentralized multi-agent cooperative learning is a practical task due to the partially observed setting both in training and execution. Every agent learns to cooperate without access to the observations and policies of others. However, the decentralized training of multi-agent is of great difficulty due to non-stationarity, especially when other agents' policies are also in learning during training. To overcome this, we propose to learn a dynamic policy belief for each agent to predict the current policies of other agents and accordingly condition the policy of its own. To quickly adapt to the development of others' policies, we introduce a historical context to learn the belief inference according to a few recent action histories of other agents and a latent variational inference to model their policies by a learned distribution. We evaluate our method on the StarCraft II micro management task (SMAC) and demonstrate its superior performance in the decentralized training settings and comparable results with the state-of-the-art CTDE methods.

Weakly Supervised Object Localization (WSOL) aims to localize objects using only image-level labels while ensuring competitive classification performance. However, previous efforts have prioritized localization over classification accuracy in discriminative features, in which low-level information is neglected. We argue that low-level image representations, such as edges, color, texture, and motions are crucial for accurate detection. That is, using such information further achieves more refined localization, which can be used to promote classification accuracy. In this paper, we propose a unified framework that simultaneously improves localization and classification accuracy, termed as LocLoc (Low-level Cues and Local-area Guides). It leverages low-level image cues to explore global and local representations for accurate localization and classification. Specifically, we introduce a GrabCut-Enhanced Generator (GEG) to learn global semantic representations for localization based on graph cuts to enhance low-level information based on long-range dependencies captured by the transformer. We further design a Local Feature Digging Module (LFDM) that utilizes low-level cues to guide the learning route of local feature representations for accurate classification. Extensive experiments demonstrate the effectiveness of LocLoc with 84.4%(↑5.2%) Top-1 Loc., 85.8% Top-1 Cls. on CUB-200-2011 and 57.6% (↑1.5%) Top-1 Loc., 78.6% Top-1Cls. on ILSVRC 2012, indicating that our method achieves competitive performance with a large margin compared to previous approaches. Code and models are available at https://github.com/Cliffia123/LocLoc.

Audio-visual zero-shot learning (ZSL) has attracted board attention, as it could classify video data from classes that are not observed during training. However, most of the existing methods are restricted to background scene bias and fewer motion details by employing a single-stream network to process scenes and motion information as a unified entity. In this paper, we address this challenge by proposing a novel dual-stream architecture Motion-Decoupled Spiking Transformer (MDFT) to explicitly decouple the contextual semantic information and highly sparsity dynamic motion information. Specifically, The Recurrent Joint Learning Unit (RJLU) could extract contextual semantic information effectively and understand the environment in which actions occur by capturing joint knowledge between different modalities. By converting RGB images to events, our approach effectively captures motion information while mitigating the influence of background scene biases, leading to more accurate classification results. We utilize the inherent strengths of Spiking Neural Networks (SNNs) to process highly sparsity event data efficiently. Additionally, we introduce a Discrepancy Analysis Block (DAB) to model the audio motion features. To enhance the efficiency of SNNs in extracting dynamic temporal and motion information, we dynamically adjust the threshold of Leaky Integrate-and-Fire (LIF) neurons based on the statistical cues of global motion and contextual semantic information. Our experiments demonstrate the effectiveness of MDFT, which consistently outperforms state-of-the-art methods across mainstream benchmarks. Moreover, we find that motion information serves as a powerful regularization for video networks, where using it improves the accuracy of HM and ZSL by 19.1% and 38.4%, respectively.

Event cameras offer the advantages of low latency, high temporal resolution and HDR compared to conventional cameras. Due to the asynchronous and sparse nature of events, many existing algorithms cannot be directly applied, necessitating the reconstruction of intensity frames. However, existing reconstruction methods often result in artifacts and edge blurring due to noise and event accumulation. In this paper, we argue that the key to event-based image reconstruction is to enhance the edge information of objects and restore the artifacts in the reconstructed images. To explain, edge information is one of the most important features in the event stream, providing information on the shape and contour of objects. Considering the extraordinary capabilities of Denoising Diffusion Probabilistic Models (DDPMs) in image generation, reconstruction, and restoration, we propose a new framework which incorporate it into the reconstruction pipeline to obtain high-quality results which effectively remove artifacts and blur in reconstructed images. Specifically, we first extract edge information from the event stream using the proposed event-based denoising method. It employs the contrast maximization framework to remove noise from the event stream and extract clear object edge information. And then, the edge information is further adopted to our diffusion model, which is used to enhance the edges of objects in the reconstructed images, thus improving the restoration effect. Experimental results show that our method achieves significant improvements in the mean squared error (MSE), the structural similarity (SSIM), and the perceptual similarity (LPIPS) metrics, with average improvements of 40%, 15%, and 25%, respectively, compared to previous state-of-the-art models, and has good generalization performance.

Human-centric visual analysis is a fundamental task for many multimedia and computer vision applications, such as self-driving, multimedia retrieval, and augmented reality, etc. Based on our recent research efforts on fine-grained human visual analysis, we develop a robust and efficient human-centric visual analysis system named as HumVis. HumVis is built on a simple yet efficient contextual instance decoupling (CID) module, which can effectively separate different persons in an input image and output corresponding person structure information for visual analysis. Based on CID, HumVis achieves accurate multi-person pose estimation, multi-person foreground segmentation, multi-person part segmentation and 3D human mesh recovery for user-uploaded images/videos and support live stream presentation.

Pre-trained language models (PLMs) have demonstrated strong performance in sequential recommendation (SR), which are utilized to extract general knowledge. However, existing methods still lack domain knowledge and struggle to capture users' fine-grained preferences. Meanwhile, many traditional SR methods improve this issue by integrating side information while suffering from information loss. To summarize, we believe that a good recommendation system should utilize both general and domain knowledge simultaneously. Therefore, we introduce an external knowledge base and propose Knowledge Prompt-tuning for Sequential Recommendation (KP4SR). Specifically, we construct a set of relationship templates and transform a structured knowledge graph (KG) into knowledge prompts to solve the problem of the semantic gap. However, knowledge prompts disrupt the original data structure and introduce a significant amount of noise. We further construct a knowledge tree and propose a knowledge tree mask, which restores the data structure in a mask matrix form, thus mitigating the noise problem. We evaluate KP4SR on three real-world datasets, and experimental results show that our approach outperforms state-of-the-art methods on multiple evaluation metrics. Specifically, compared with PLM-based methods, our method improves NDCG@5 and HR@5 by 40.65% and 36.42% on the books dataset, 11.17% and 11.47% on the music dataset, and 22.17% and 19.14% on the movies dataset, respectively. Our code is publicly available at the link: https://github.com/zhaijianyang/KP4SR.

Vanilla spiking neurons in Spiking Neural Networks (SNNs) use charge-fire-reset neuronal dynamics, which can only be simulated serially and can hardly learn long-time dependencies. We find that when removing reset, the neuronal dynamics can be reformulated in a non-iterative form and parallelized. By rewriting neuronal dynamics without reset to a general formulation, we propose the Parallel Spiking Neuron (PSN), which generates hidden states that are independent of their predecessors, resulting in parallelizable neuronal dynamics and extremely high simulation speed. The weights of inputs in the PSN are fully connected, which maximizes the utilization of temporal information. To avoid the use of future inputs for step-by-step inference, the weights of the PSN can be masked, resulting in the masked PSN. By sharing weights across time-steps based on the masked PSN, the sliding PSN is proposed to handle sequences of varying lengths. We evaluate the PSN family on simulation speed and temporal/static data classification, and the results show the overwhelming advantage of the PSN family in efficiency and accuracy. To the best of our knowledge, this is the first study about parallelizing spiking neurons and can be a cornerstone for the spiking deep learning research. Our codes are available at https://github.com/fangwei123456/Parallel-Spiking-Neuron.

Integrating RGB frames with alternative modality inputs is gaining increasing traction in many vision-based reinforcement learning (RL) applications. Existing multi-modal vision-based RL methods usually follow a Global Value Estimation (GVE) pipeline, which uses a fused modality feature to obtain a unified global environmental description. However, such a feature-level fusion paradigm with a single critic may fall short in policy learning as it tends to overlook the distinct values of each modality. To remedy this, this paper proposes a Local modality-customized Value Estimation (LVE) paradigm, which dynamically estimates the contribution and adjusts the importance weight of each modality from a value-level perspective. Furthermore, a task-contextual re-fusion process is developed to achieve a task-level re-balance of estimations from both feature and value levels. To this end, a Hierarchical Adaptive Value Estimation (HAVE) framework is formed, which adaptively coordinates the contributions of individual modalities as well as their collective efficacy. Agents trained by HAVE are able to exploit the unique characteristics of various modalities while capturing their intricate interactions, achieving substantially improved performance. We specifically highlight the potency of our approach within the challenging landscape of autonomous driving, utilizing the CARLA benchmark with neuromorphic event and depth data to demonstrate HAVE's capability and the effectiveness of its distinct components.

Spiking Neural Networks (SNNs) provide an energy-efficient deep learning option due to their unique spike-based event-driven (i.e., spike-driven) paradigm. In this paper, we incorporate the spike-driven paradigm into Transformer by the proposed Spike-driven Transformer with four unique properties: (1) Event-driven, no calculation is triggered when the input of Transformer is zero; (2) Binary spike communication, all matrix multiplications associated with the spike matrix can be transformed into sparse additions; (3) Self-attention with linear complexity at both token and channel dimensions; (4) The operations between spike-form Query, Key, and Value are mask and addition. Together, there are only sparse addition operations in the Spike-driven Transformer. To this end, we design a novel Spike-Driven Self-Attention (SDSA), which exploits only mask and addition operations without any multiplication, and thus having up to $87.2\times$ lower computation energy than vanilla self-attention. Especially in SDSA, the matrix multiplication between Query, Key, and Value is designed as the mask operation. In addition, we rearrange all residual connections in the vanilla Transformer before the activation functions to ensure that all neurons transmit binary spike signals. It is shown that the Spike-driven Transformer can achieve 77.1\% top-1 accuracy on ImageNet-1K, which is the state-of-the-art result in the SNN field.

Neuromorphic vision sensor is a new bio-inspired imaging paradigm emerged in recent years. It uses the asynchronous spike signals instead of the traditional frame-based manner to achieve ultra-high speed sampling. Unlike the dynamic vision sensor (DVS) that perceives movement by imitating the retinal periphery, the spike camera was developed recently to perceive fine textures by simulating a small retinal region called the fovea. For this new type of neuromorphic camera, how to reconstruct ultra-high speed visual images from spike data becomes an important yet challenging issue in visual scene perception, analysis, and recognition applications. In this paper, a bio-inspired visual reconstruction framework for the spike camera is proposed for the first time. Its core idea is to use the biologically inspired adaptive adjustment mechanisms, combined with the spatiotemporal spike information extracted by the proposed model, to reconstruct the full texture of natural scenes in an ultra-high temporal resolution. Specifically, the proposed model consists of a motion local excitation layer, a spike refining layer and a visual reconstruction layer motivated by the bio-realistic leaky integrate-and-fire (LIF) neurons and synapse connection with spike-timing dependent plasticity (STDP) rule. To evaluate the performance, a spike dataset was constructed for normal and high-speed scenes in real-world recorded by the spike camera. The experimental results show that the proposed approach can reconstruct the visual images with 40,000 frames per second in both normal and high-speed scenes, while achieving high dynamic range and high image quality.

DAVIS camera, streaming two complementary sensing modalities of asynchronous events and frames, has gradually been used to address major object detection challenges (e.g., fast motion blur and low-light). However, how to effectively leverage rich temporal cues and fuse two heterogeneous visual streams remains a challenging endeavor. To address this challenge, we propose a novel streaming object detector with Transformer, namely SODFormer, which first integrates events and frames to continuously detect objects in an asynchronous manner. Technically, we first build a large-scale multimodal neuromorphic object detection dataset (i.e., PKU-DAVIS-SOD) over 1080.1 k manual labels. Then, we design a spatiotemporal Transformer architecture to detect objects via an end-to-end sequence prediction problem, where the novel temporal Transformer module leverages rich temporal cues from two visual streams to improve the detection performance. Finally, an asynchronous attention-based fusion module is proposed to integrate two heterogeneous sensing modalities and take complementary advantages from each end, which can be queried at any time to locate objects and break through the limited output frequency from synchronized frame-based fusion strategies. The results show that the proposed SODFormer outperforms four state-of-the-art methods and our eight baselines by a significant margin. We also show that our unifying framework works well even in cases where the conventional frame-based camera fails, e.g., high-speed motion and low-light conditions. Our dataset and code can be available at https://github.com/dianzl/SODFormer.

Though network pruning receives popularity in reducing the complexity of convolutional neural networks (CNNs), it remains an open issue to concurrently maintain model accuracy as well as achieve significant speedups on general CPUs. In this paper, we propose a novel 1×N pruning pattern to break this limitation. In particular, consecutive N output kernels with the same input channel index are grouped into one block, which serves as a basic pruning granularity of our pruning pattern. Our 1×N pattern prunes these blocks considered unimportant. We also provide a workflow of filter rearrangement that first rearranges the weight matrix in the output channel dimension to derive more influential blocks for accuracy improvements and then applies similar rearrangement to the next-layer weights in the input channel dimension to ensure correct convolutional operations. Moreover, the output computation after our 1×N pruning can be realized via a parallelized block-wise vectorized operation, leading to significant speedups on general CPUs. The efficacy of our pruning pattern is proved with experiments on ILSVRC-2012. For example, given the pruning rate of 50% and N=4, our pattern obtains about 3.0% improvements over filter pruning in the top-1 accuracy of MobileNet-V2. Meanwhile, it obtains 56.04ms inference savings on Cortex-A7 CPU over weight pruning. Our project is made available at https://github.com/lmbxmu/1xN.

Unsupervised domain adaption has been widely adopted in tasks with scarce annotated data. Unfortunately, mapping the target-domain distribution to the source-domain unconditionally may distort the essential structural information of the target-domain data, leading to inferior performance. To address this issue, we first propose to introduce active sample selection to assist domain adaptation regarding the semantic segmentation task. By innovatively adopting multiple anchors instead of a single centroid, both source and target domains can be better characterized as multimodal distributions, in which way more complementary and informative samples are selected from the target domain. With only a little workload to manually annotate these active samples, the distortion of the target-domain distribution can be effectively alleviated, achieving a large performance gain. In addition, a powerful semi-supervised domain adaptation strategy is proposed to alleviate the long-tail distribution problem and further improve the segmentation performance. Extensive experiments are conducted on public datasets, and the results demonstrate that the proposed approach outperforms state-of-the-art methods by large margins and achieves similar performance to the fully-supervised upperbound, i.e., 71.4% mIoU on GTA5 and 71.8% mIoU on SYNTHIA. The effectiveness of each component is also verified by thorough ablation studies.

Brain-inspired spiking neural networks (SNNs) are becoming a promising energy-efficient alternative to traditional artificial neural networks (ANNs). However, the performance gap between SNNs and ANNs has been a significant hindrance to deploying SNNs ubiquitously. To leverage the full potential of SNNs, in this paper we study the attention mechanisms, which can help human focus on important information. We present our idea of attention in SNNs with a multi-dimensional attention module, which infers attention weights along the temporal, channel, as well as spatial dimension separately or simultaneously. Based on the existing neuroscience theories, we exploit the attention weights to optimize membrane potentials, which in turn regulate the spiking response. Extensive experimental results on event-based action recognition and image classification datasets demonstrate that attention facilitates vanilla SNNs to achieve sparser spiking firing, better performance, and energy efficiency concurrently. In particular, we achieve top-1 accuracy of 75.92% and 77.08% on ImageNet-1 K with single/4-step Res-SNN-104, which are state-of-the-art results in SNNs. Compared with counterpart Res-ANN-104, the performance gap becomes -0.95/+0.21 percent and the energy efficiency is 31.8×/7.4×. To analyze the effectiveness of attention SNNs, we theoretically prove that the spiking degradation or the gradient vanishing, which usually holds in general SNNs, can be resolved by introducing the block dynamical isometry theory. We also analyze the efficiency of attention SNNs based on our proposed spiking response visualization method. Our work lights up SNN's potential as a general backbone to support various applications in the field of SNN research, with a great balance between effectiveness and energy efficiency.

Human part segmentation is a crucial but challenging task in computer vision. Recent works have achieved progress with the help of pixel-wise annotations. However, annotating pixel-wise masks especially at part-level is a tedious and labor-intensive procedure. To overcome this problem, we propose a part evolution framework to learn reliable predictions from weak pose annotations, which are much easier to collect. Our framework is composed of two essential modules: the first part adaptation module is designed to learn the deep prior knowledge from three related tasks, i.e., pose estimation, part-level and object-level segmentation; the second module is the part evolution module, which refines the part priors from deep predictions with the boundary-aware optimization algorithm. These two modules are conducted iteratively to evolve pose keypoint annotations into reliable part priors. Experimental evidence shows that our weakly-supervised approach generates comparable results with the state-of-the-art strongly-supervised methods on public benchmarks, and also validates the potential of notable improvements when combining weak labels with existing part segmentation masks.

Few-shot learning aims to recognize novel queries with limited support samples by learning from base knowledge. Recent progress in this setting assumes that the base knowledge and novel query samples are distributed in the same domains, which are usually infeasible for realistic applications. Toward this issue, we propose to address the cross-domain few-shot learning problem where only extremely few samples are available in target domains. Under this realistic setting, we focus on the fast adaptation capability of meta-learners by proposing an effective dual adaptive representation alignment approach. In our approach, a prototypical feature alignment is first proposed to recalibrate support instances as prototypes and reproject these prototypes with a differentiable closed-form solution. Therefore feature spaces of learned knowledge can be adaptively transformed to query spaces by the cross-instance and cross-prototype relations. Besides the feature alignment, we further present a normalized distribution alignment module, which exploits prior statistics of query samples for solving the covariant shifts among the support and query samples. With these two modules, a progressive meta-learning framework is constructed to perform the fast adaptation with extremely few-shot samples while maintaining its generalization capabilities. Experimental evidence demonstrates our approach achieves new state-of-the-art results on 4 CDFSL benchmarks and 4 fine-grained cross-domain benchmarks.

Recently, remarkable progress has been made in learned image compression (LIC), in which nonlinear transforms (NTs) play a crucial role. Although there are many NT methods for improving the rate distortion performance, all the existing methods sacrifice the computational complexity and the number of parameters of the transformation. This paper provides a fundamental novel viewpoint on nonlinear transforms from a communication perspective, and shows how this idea can be extended to design efficient NT methods. In particular, the nonlinear transforms are inferred as signal modulation modules. Under this extrapolation, the current NTs are generalized as amplitude modulation that only varies the amplitude of the carrier wave. Therefore, a nonlinear modulation-like transform (NMLT) which varies the phase angle of the carrier is proposed. Moreover, this concept is extended by introducing In-phase/Quadrature (IQ) modulation, which is a boosting technique in communication field, in order to enhance NMLT. Furthermore, the Bit-interleaved technique in communication is used to guide the optimization of NTML with IQ. The experimental results on different datasets and backbone architectures verify the efficiency and robustness of the proposed methods. For example, when backbone architecture is hyperprior model, our method achieves 19.37% BD-rate reduction over GDN on the Kodak dataset. In addition, our method with channel wise autoregressive model leads to the state-of-the-art rate-distortion performance.

Most of the image-text retrieval methods carry out accurate results using fine-grained features for feature alignment. However, extracting the robustness features while maintaining the retrieval accuracy in wireless communication is still a challenge, especially with channel noises and limited transmission bandwidth. Inspired by spike signals of neurons in the human brain, we propose the neuron-based spiking transmission and reasoning network (NSTRN). In this way, the features are compressed into compacted efficient representations. In NSTRN, we construct the feature sender based on spiking activation function to selectively encode only important information in images and sentences into binary codes, and reduce the transmission cost. Moreover, the feature receiver is designed as a recurrent architecture and applies both temporal attention and global attention blocks to memorize long-term information. Finally, to compensate for the loss of visual concepts in transmission, we use the global textual features as coefficients to guide the formation of visual features in the training stage. The traditional CNN-based joint source-channel coding model outputs float-point encoded features, which requires additional quantization steps to convert features into binary bitstreams in the practical wireless communication system. Instead, the spiking neural networks (SNNs) directly use binary spike trains to reduce the computation complexity caused by the quantization steps. More importantly, SNNs can naturally encode the asynchronous event streams and inhibit the discrete noisy events to extract robust information. Even with binary bitstreams, NSTRN shows effectiveness compared with the state-of-the-art image-text retrieval methods. In the wireless communication scenario, NSTRN not only reduces the transmission bandwidth but also alleviates the “cliff effect” to a certain extent in the traditional separate encoding methods. To the best of our knowledge, this is the first work using SNNs on robust image-text retrieval.

The sensitivity of deep neural networks to compressed images hinders their usage in many real applications, which means classification networks may fail just after taking a screenshot and saving it as a compressed file. In this paper, we argue that neglected disposable coding parameters stored in compressed files could be picked up to reduce the sensitivity of deep neural networks to compressed images. Specifically, we resort to using one of the representative parameters, quantization steps, to facilitate image classification. Firstly, based on quantization steps, we propose a novel quantization aware confidence (QAC), which is utilized as sample weights to reduce the influence of quantization on network training. Secondly, we utilize quantization steps to alleviate the variance of feature distributions, where a quantization aware batch normalization (QABN) is proposed to replace batch normalization of classification networks. Extensive experiments show that the proposed method significantly improves the performance of classification networks on CIFAR-10, CIFAR-100, and ImageNet.

Spike camera is a new type of bio-inspired vision sensor, each pixel of which perceives the brightness of the scene independently, and finally outputs 3-dimensional spatiotemporal spike streams. To bridge the spike camera and traditional frame-based vision, there is some works to reconstruct spike streams into regular images. However, the low spatial resolution ($400\times 250$) of the spike camera limits the quality of the reconstructed images. Thus, it is meaningful to explore a super-resolution reconstruction for spike streams. In this paper, we propose an end-to-end network to reconstruct high-resolution images from low-resolution spike streams. To utilize more spatiotemporal features of spike streams, our network adopts a multi-level features learning mechanism, including intra-stream feature extraction by spike encoder, inter-stream dependencies extraction based on optical flow module, and joint features learning via spike-based iterative projection. Experimental results demonstrate that our network is superior to the combination of state-of-the-art intensity image reconstruction methods and super-resolution networks on simulated and real datasets.

Complicated temporal patterns can provide important information for accurate time series forecasting. Existing long short-term memory (LSTM) model with attention mechanism have achieved significant performance. However, the exponential decay of long-term memory of LSTM has not be resolved yet in these efforts, remaining a longstanding open problem in recurrent nature. This problem exhibits a bottleneck which restricts the performance of existing studies. Recently, spiking neural networks (SNNs) have shown high efficiency in capturing temporal patterns via the surrogate gradient (SG) method to resolve this issue. However, the concept-drift environment makes it impossible to pre-set the variance into the standard SG method due to time-varying data distribution. In this paper, we propose a novel adaptive and hybrid spiking (AHS) module embedded LSTM, collaborating with two attention mechanisms (called HSN-LSTM) to resolve above-mentioned problems. First, the AHS module is analyzed theoretically can remain long-term memory. Moreover, our smooth SG method avoids pre-setting of variance, which is not sensitive in the above scenarios. Besides, we use the negative log-likelihood function to adjust the attention score for alleviating the negative impact from the concept-drift. Experiment results show the HSN-LSTM outperformed the state-of-the-art models on several multivariate time series datasets.

Traffic signal control aims to coordinate traffic signals across intersections to improve the traffic efficiency of a district or a city. Deep reinforcement learning (RL) has been applied to traffic signal control recently and demonstrated promising performance where each traffic signal is regarded as an agent. However, there are still several challenges that may limit its large-scale application in the real world. On the one hand, the policy of the current traffic signal is often heavily influenced by its neighbor agents, and the coordination between the agent and its neighbors needs to be considered. Hence, the control of a road network composed of multiple traffic signals is naturally modeled as a multi-agent system, and all agents’ policies need to be optimized simultaneously. On the other hand, once the policy function is conditioned on not only the current agent's observation but also the neighbors’, the policy function would be closely related to the training scenario and cause poor generalizability because the agents in various scenarios often have heterogeneous neighbors. To make the policy learned from a training scenario generalizable to new unseen scenarios, a novel Meta Variationally Intrinsic Motivated (MetaVIM) RL method is proposed to learn the decentralized policy for each intersection that considers neighbor information in a latent way. Specifically, we formulate the policy learning as a meta-learning problem over a set of related tasks, where each task corresponds to traffic signal control at an intersection whose neighbors are regarded as the unobserved part of the state. Then, a learned latent variable is introduced to represent the task's specific information and is further brought into the policy for learning. In addition, to make the policy learning stable, a novel intrinsic reward is designed to encourage each agent's received rewards and observation transition to be predictable only conditioned on its own history. Extensive experiments conducted on CityFlow demonstrate that the proposed method substantially outperforms existing approaches and shows superior generalizability.

Semi-supervised learning (SSL) methods show their powerful performance to deal with the issue of data shortage in the field of medical image segmentation. However, existing SSL methods still suffer from the problem of unreliable predictions on unannotated data due to the lack of manual annotations for them. In this paper, we propose an unreliability-diluted consistency training (UDiCT) mechanism to dilute the unreliability in SSL by assembling reliable annotated data into unreliable unannotated data. Specifically, we first propose an uncertainty-based data pairing module to pair annotated data with unannotated data based on a complementary uncertainty pairing rule, which avoids two hard samples being paired off. Secondly, we develop SwapMix, a mixed sample data augmentation method, to integrate annotated data into unannotated data for training our model in a low-unreliability manner. Finally, UDiCT is trained by minimizing a supervised loss and an unreliability-diluted consistency loss, which makes our model robust to diverse backgrounds. Extensive experiments on three chest CT datasets show the effectiveness of our method for semi-supervised CT lesion segmentation.

Many RGB-T trackers attempt to attain robust feature representation by utilizing an adaptive weighting scheme (or attention mechanism). Different from these works, we propose a new dynamic modality-aware filter generation module (named MFGNet) to boost the message communication between visible and thermal data by adaptively adjusting the convolutional kernels for various input images in practical tracking. Given the image pairs as input, we first encode their features with the backbone network. Then, we concatenate these feature maps and generate dynamic modality-aware filters with two independent networks. The visible and thermal filters will be used to conduct a dynamic convolutional operation on their corresponding input feature maps respectively. Inspired by residual connection, both the generated visible and thermal feature maps will be summarized with input feature maps. The augmented feature maps will be fed into the RoI align module to generate instance-level features for subsequent classification. To address issues caused by heavy occlusion, fast motion and out-of-view, we propose to conduct a joint local and global search by exploiting a new direction-aware target driven attention mechanism. The spatial and temporal recurrent neural network is used to capture the direction-aware context for accurate global attention prediction. Extensive experiments on three large-scale RGB-T tracking benchmark datasets validated the effectiveness of our proposed algorithm.

Event cameras as bioinspired vision sensors have shown great advantages in high dynamic range and high temporal resolution in vision tasks. Asynchronous spikes from event cameras can be depicted using the marked spatiotemporal point processes (MSTPPs). However, how to measure the distance between asynchronous spikes in the MSTPPs still remains an open issue. To address this problem, we propose a general asynchronous spatiotemporal spike metric considering both spatiotemporal structural properties and polarity attributes for event cameras. Technically, the conditional probability density function is first introduced to describe the spatiotemporal distribution and polarity prior in the MSTPPs. Besides, a spatiotemporal Gaussian kernel is defined to capture the spatiotemporal structure, which transforms discrete spikes into the continuous function in a reproducing kernel Hilbert space (RKHS). Finally, the distance between asynchronous spikes can be quantified by the inner product in the RKHS. The experimental results demonstrate that the proposed approach outperforms the state-of-the-art methods and achieves significant improvement in computational efficiency. Especially, it is able to better depict the changes involving spatiotemporal structural properties and polarity attributes.

Existing online knowledge distillation approaches either adopt the student with the best performance or construct an ensemble model for better holistic performance. However, the former strategy ignores other students' information, while the latter increases the computational complexity during deployment. In this article, we propose a novel method for online knowledge distillation, termed feature fusion and self-distillation (FFSD), which comprises two key components: FFSD, toward solving the above problems in a unified framework. Different from previous works, where all students are treated equally, the proposed FFSD splits them into a leader student set and a common student set. Then, the feature fusion module converts the concatenation of feature maps from all common students into a fused feature map. The fused representation is used to assist the learning of the leader student. To enable the leader student to absorb more diverse information, we design an enhancement strategy to increase the diversity among students. Besides, a self-distillation module is adopted to convert the feature map of deeper layers into a shallower one. Then, the shallower layers are encouraged to mimic the transformed feature maps of the deeper layers, which helps the students to generalize better. After training, we simply adopt the leader student, which achieves superior performance, over the common students, without increasing the storage or inference cost. Extensive experiments on CIFAR-100 and ImageNet demonstrate the superiority of our FFSD over existing works. The code is available at https://github.com/SJLeo/FFSD.

Channel pruning has been long studied to compress convolutional neural networks (CNNs), which significantly reduces the overall computation. Prior works implement channel pruning in an unexplainable manner, which tends to reduce the final classification errors while failing to consider the internal influence of each channel. In this article, we conduct channel pruning in a white box. Through deep visualization of feature maps activated by different channels, we observe that different channels have a varying contribution to different categories in image classification. Inspired by this, we choose to preserve channels contributing to most categories. Specifically, to model the contribution of each channel to differentiating categories, we develop a class-wise mask for each channel, implemented in a dynamic training manner with respect to the input image's category. On the basis of the learned class-wise mask, we perform a global voting mechanism to remove channels with less category discrimination. Lastly, a fine-tuning process is conducted to recover the performance of the pruned model. To our best knowledge, it is the first time that CNN interpretability theory is considered to guide channel pruning. Extensive experiments on representative image classification tasks demonstrate the superiority of our White-Box over many state-of-the-arts (SOTAs). For instance, on CIFAR-10, it reduces 65.23% floating point operations per seconds (FLOPs) with even 0.62% accuracy improvement for ResNet-110. On ILSVRC-2012, White-Box achieves a 45.6% FLOP reduction with only a small loss of 0.83% in the top-1 accuracy for ResNet-50. Code is available at https://github.com/zyxxmu/White-Box.

Due to high-speed motion blur and challenging illumination, conventional frame-based cameras have encountered an important challenge in object detection tasks. Neuromorphic cameras that output asynchronous visual streams instead of intensity frames, by taking the advantage of high temporal resolution and high dynamic range, have brought a new perspective to address the challenge. In this paper, we propose a novel problem setting, retinomorphic object detection, which is the first trial that integrates foveal-like and peripheral-like visual streams. Technically, we first build a large-scale multimodal neuromorphic object detection dataset (i.e., PKU-Vidar-DVS) over 215.5k spatio-temporal synchronized labels. Then, we design temporal aggregation representations to preserve the spatio-temporal information from asynchronous visual streams. Finally, we present a novel bio-inspired unifying framework to fuse two sensing modalities via a dynamic interaction mechanism. Our experimental evaluation shows that our approach has significant improvements over the state-of-the-art methods with the single-modality, especially in high-speed motion and low-light scenarios. We hope that our work will attract further research into this newly identified, yet crucial research direction. Our dataset can be available at https://www.pkuml.org/resources/pku-vidar-dvs.html.

Drug-target interaction(DTI) prediction is one of the most important topics in drug design and drug development, and deep learning approaches have achieved state-of-the-art performance in this field. However, the current methods are difficult to successfully combine the local and global features of drug molecules and protein sequences, while ignoring the modeling of complicated interaction mechanisms, which leads to a certain limitation of prediction performance. To overcome this barrier, we propose an end-to-end method based on Convolutional Neural Network (CNN) and Transformer to predict DTI problems, named ConformerDTI. The CNN and Transformer branches extract features from the simplified molecular input line entry system (SMILES) string of drugs and the amino acid sequence of proteins, respectively. The local and global features are coupled by the mutual transfer of the two branches through cross attention. Decoupling of local and global features in parallel leverages CNN’s power in extracting local features as well as the efficiency of Transformer at global processing. I n addition, ConformerDTI exploits the convolutional interaction network to model the interaction mechanism, both drugs and targets are convoluted by dynamic filters generated based on each other. Experimental results demonstrate that our model has better prediction performance than the most advanced deep learning methods on three different datasets. Furthermore, this performance improvement was validated by ablation experiments.

Neuromorphic vision sensor is a new bio-inspired imaging paradigm that reports asynchronous, continuously perpixel brightness changes called ‘events’ with high temporal resolution and high dynamic range. So far, the event-based image reconstruction methods are based on artificial neural networks (ANN) or hand-crafted spatiotemporal smoothing techniques. In this paper, we first implement the image reconstruction work via deep spiking neural network (SNN) architecture. As the bio-inspired neural networks, SNNs operating with asynchronous binary spikes distributed over time, can potentially lead to greater computational efficiency on event-driven hardware. We propose a novel Event-based Video reconstruction framework based on a fully Spiking Neural Network (EVSNN), which utilizes Leaky-Integrate-and-Fire (LIF) neuron and Membrane Potential (MP) neuron. We find that the spiking neurons have the potential to store useful temporal information (memory) to complete such time-dependent tasks. Further-more, to better utilize the temporal information, we propose a hybrid potential-assisted framework (PAEVSNN) using the membrane potential of spiking neuron. The proposed neuron is referred as Adaptive Membrane Potential (AMP) neuron, which adaptively updates the membrane potential according to the input spikes. The experimental results demonstrate that our models achieve comparable performance to ANN-based models on IJRR, MVSEC, and HQF datasets. The energy consumptions of EVSNN and PAEVSNN are$19.36\times$and$7.75\times$more computationally ef-ficient than their ANN architectures, respectively. The code and pretrained model are available at https://sites.google.com/view/evsnn.

Semantic patterns offine-grained objects are determined by subtle appearance difference of local parts, which thus inspires a number of part-based methods. However, due to uncontrollable object poses in images, distinctive de-tails carried by local regions can be spatially distributed or even self-occluded, leading to a large variation on ob-ject representation. For discounting pose variations, this paper proposes to learn a novel graph based object rep-resentation to reveal a global configuration of local parts for self-supervised pose alignment across classes, which is employed as an auxiliary feature regularization on a deep representation learning network. Moreover, a coarse-to-fine supervision together with the proposed pose-insensitive constraint on shallow-to-deep sub-networks encourages discriminative features in a curriculum learning manner. We evaluate our method on three popular fine-grained ob-ject classification benchmarks, consistently achieving the state-of-the-art performance. Source codes are available at https://github.com/yangxhll/P2P-Net.

Image deraining is a typical low-level image restoration task, which aims at decomposing the rainy image into two distinguishable layers: clean image layer and rain layer. Most of the existing learning-based deraining methods are supervisedly trained on synthetic rainy-clean pairs. The domain gap between the synthetic and real rains makes them less generalized to different real rainy scenes. Moreover, the existing methods mainly utilize the property of the two layers independently, while few of them have considered the mutually exclusive relationship between the two layers. In this work, we propose a novel non-local contrastive learning (NLCL) method for unsupervised image deraining. Consequently, we not only utilize the intrinsic self-similarity property within samples, but also the mutually exclusive property between the two layers, so as to better differ the rain layer from the clean image. Specifically, the non-local self-similarity image layer patches as the positives are pulled together and similar rain layer patches as the negatives are pushed away. Thus the similar positive/negative samples that are close in the original space benefit us to enrich more discriminative representation. Apart from the self-similarity sampling strategy, we analyze how to choose an appropriate feature encoder in NLCL. Extensive experiments on different real rainy datasets demonstrate that the proposed method obtains state-of-the-art performance in real deraining.

Performance evaluation strategy is one of the most important factors that determine the effectiveness and efficiency in Neural Architecture Search (NAS). Existing strategies, such as employing standard training or performance predictor, often suffer from high computational complexity and low generality. To address this issue, we propose to rank architectures by Representation Mutual Information (RMI). Specifically, given an arbitrary architecture that has decent accuracy, architectures that have high RMI with it always yield good accuracies. As an accurate performance indicator to facilitate NAS, RMI not only generalizes well to different search spaces, but is also efficient enough to evaluate architectures using only one batch of data. Building upon RMI, we further propose a new search algorithm termed RMI-NAS, facilitating with a theorem to guarantee the global optimal of the searched architecture. In particular, RMI-NAS first randomly samples architectures from the search space, which are then effectively classified as positive or negative samples by RMI. We then use these samples to train a random forest to explore new regions, while keeping track of the distribution of positive architectures. When the sample size is sufficient, the architecture with the largest probability from the aforementioned distribution is selected, which is theoretically proved to be the optimal solution. The architectures searched by our method achieve remarkable top-1 accuracies with the magnitude times faster search process. Besides, RMI-NAS also generalizes to different datasets and search spaces. Our code has been made available at https://git.openi.org.cn/PCL_AutoML/XNAS.

Learning to synthesize data has emerged as a promising direction in zero-shot quantization (ZSQ), which represents neural networks by low-bit integer without accessing any of the real data. In this paper, we observe an interesting phenomenon of intra-class heterogeneity in real data and show that existing methods fail to retain this property in their synthetic images, which causes a limited performance increase. To address this issue, we propose a novel zero-shot quantization method referred to as IntraQ. First, we propose a local object reinforcement that locates the target objects at different scales and positions of the synthetic images. Second, we introduce a marginal distance constraint to form class-related features distributed in a coarse area. Lastly, we devise a soft inception loss which injects a soft prior label to prevent the synthetic images from being over-fitting to a fixed object. Our IntraQ is demonstrated to well retain the intra-class heterogeneity in the synthetic images and also observed to perform state-of-the-art. For example, compared to the advanced ZSQ, our IntraQ obtains 9.17% increase of the top-1 accuracy on ImageNet when all layers of MobileNetV1 are quantized to 4-bit. Code is at https://github.com/zysxmu/IntraQ

Recently, Vision Transformer (ViT) has achieved remarkable success in several computer vision tasks. The progresses are highly relevant to the architecture design, then it is worthwhile to propose Transformer Architecture Search (TAS) to search for better ViTs automatically. However, current TAS methods are time-consuming and existing zero-cost proxies in CNN do not generalize well to the ViT search space according to our experimental observations. In this paper, for the first time, we investigate how to conduct TAS in a training-free manner and devise an effective training-free TAS (TF-TAS) scheme. Firstly, we observe that the properties of multi-head self-attention (MSA) and multi-layer perceptron (MLP) in ViTs are quite different and that the synaptic diversity of MSA affects the performance notably. Secondly, based on the observation, we devise a modular strategy in TF-TAS that evaluates and ranks ViT architectures from two theoretical perspectives: synaptic diversity and synaptic saliency, termed as DSS-indicator. With DSS-indicator, evaluation results are strongly corre-lated with the test accuracies of ViT models. Experimental results demonstrate that our TF- TAS achieves a competitive performance against the state-of-the-art manually or automatically design ViT architectures, and it promotes the searching efficiency in ViT search space greatly: from about 24 GPU days to less than 0.5 GPU days. Moreover, the proposed DSS-indicator outperforms the existing cutting-edge zero-cost approaches (e.g., TE-score and NASWOT).

Power estimation is the basis of many hardware optimization strategies. However, it is still challenging to offer accurate power estimation at an early stage such as high-level synthesis (HLS). In this paper, we propose PowerGear, a graph-learning-assisted power estimation approach for FPGA HLS, which features high accuracy, efficiency and transferability. PowerGear comprises two main components: a graph construction flow and a customized graph neural network (GNN) model. Specifically, in the graph construction flow, we introduce buffer insertion, datapath merging, graph trimming and feature annotation techniques to transform HLS designs into graph-structured data, which encode both intra-operation micro-architectures and inter-operation interconnects annotated with switching activities. Furthermore, we propose a novel power-aware heterogeneous edge-centric GNN model which effectively learns heterogeneous edge semantics and structural properties of the constructed graphs via edge-centric neighborhood aggregation, and fits the formulation of dynamic power. Compared with on-board measurement, PowerGear estimates total and dynamic power for new HLS designs with errors of 3.60% and 8.81%, respectively, which outperforms the prior arts in research and the commercial product Vivado. In addition, PowerGear demonstrates a speedup of 4× over Vivado power estimator. Finally, we present a case study in which PowerGear is exploited to facilitate design space exploration for FPGA HLS, leading to a performance gain of up to 11.2%, compared with methods using state-of-the-art predictive models.

The novel working principle enables spiking cameras to capture high-speed moving objects. However, the applications of spiking cameras can be affected by many factors, such as brightness intensity, detectable distance, and the maximum speed of moving targets. Improper settings such as weak ambient brightness and too short object-camera distance, will lead to failure in the application of such cameras. To address the issue, this paper proposes a modeling algorithm that studies the detection capability of spiking cameras. The algorithm deduces the maximum detectable speed of spiking cameras corresponding to different scenario settings (e.g., brightness intensity, camera lens, and object-camera distance) based on the basic technical parameters of cameras (e.g., pixel size, spatial and temporal resolution). Thereby, the proper camera settings for various applications can be determined. Extensive experiments verify the effectiveness of the modeling algorithm. To our best knowledge, it is the first work to investigate the detection capability of spiking cameras.

Bio-inspired spike cameras, offering high temporal resolution spike streams, have brought a new perspective to address common challenges (e.g.,high-speed motion blur) in depth estimation tasks. In this paper, we propose a novel problem setting, spike-based stereo depth estimation, which is the first trail that explores an end-to-end network to learn stereo depth estimation with transformers for spike cameras, named Spike-based Stereo Depth Estimation Transformer (SSDEFormer). We first build a hybrid camera platform and provide a new stereo depth estimation dataset (i.e.,PKU-Spike-Stereo) with spatiotemporal synchronized labels. Then, we propose a novel spike representation to effectively exploit spatiotemporal information from spike streams. Finally, a transformer-based network is designed to generate dense depth maps without a fixed-disparity cost volume. Empirically, it shows that our approach is extremely effective on both synthetic and real-world datasets. The results verify that spike cameras can perform robust depth estimation even in cases where conventional cameras and event cameras fail in fast motion scenarios.

The event camera is a novel bio-inspired vision sensor. When the brightness change exceeds the preset threshold, the sensor generates events asynchronously. The number of valid events directly affects the performance of event-based tasks, such as reconstruction, detection, and recognition. However, when in low-brightness or slow-moving scenes, events are often sparse and accompanied by noise, which poses challenges for event-based tasks. To solve these challenges, we propose an event temporal up-sampling algorithm11Code: https://github.com/XIJIE-XIANG/Event-Temporal-Up-sampling to generate more effective and reliable events. The main idea of our algorithm is to generate up-sampling events on the event motion trajectory. First, we estimate the event motion trajectory by contrast maximization algorithm and then up-sampling the events by temporal point processes. Experimental results show that up-sampling events can provide more effective information and improve the performance of downstream tasks, such as improving the quality of reconstructed images and increasing the accuracy of object detection.

Spiking Neural Networks (SNNs) are considered a promising alternative to Artificial Neural Networks (ANNs) for their event-driven computing paradigm when deployed on energy-efficient neuromorphic hardware. Recently, deep SNNs have shown breathtaking performance improvement through cutting-edge training strategy and flexible structure, which also scales up the number of parameters and computational burdens in a single network. Inspired by the state transition of dendritic spines in the filopodial model of spinogenesis, we model different states of SNN weights, facilitating weight optimization for pruning. Furthermore, the pruning speed can be regulated by using different functions describing the growing threshold of state transition. We organize these techniques as a dynamic pruning algorithm based on nonlinear reparameterization mapping from spine size to SNN weights. Our approach yields sparse deep networks on the large-scale dataset (SEW ResNet18 on ImageNet) while maintaining state-of-the-art low performance loss ( 3% at 88.8% sparsity) compared to existing pruning methods on directly trained SNNs. Moreover, we find out pruning speed regulation while learning is crucial to avoiding disastrous performance degradation at the final stages of training, which may shed light on future work on SNN pruning.

Visible surface area estimation for irregular objects, one of the most fundamental and challenging topics in mathematics, supports a wide range of applications. The existing techniques usually estimate the visible surface area via mathematical modeling from 3D point clouds. However, the 3D scanner is expensive, and the corresponding evaluation method is too complex. In this paper, we propose a novel problem setting, deep learning for visible surface area estimation, which is the first trial to estimate the visible surface area for irregular objects from monocular images. Technically, we first build a novel visible surface area estimation dataset including 9099 real annotations. Then, we design a learning-based architecture to predict the visible surface area, including two core modules (i.e., the classification module and the area-bins module). The classification module is presented to predict the visible surface area distribution interval and assist network training for more accurate visible surface area estimation. Meanwhile, the area-bins module using the transformer encoder is proposed to distinguish the difference in visible surface area between irregular objects of the same category. The experimental results demonstrate that our approach can effectively estimate the visible surface area for irregular objects with various categories and sizes. We hope that this work will attract further research into this newly identified, yet crucial research direction. Our source code and data are available at \textcolormagenta \urlhttps://github.com/liuxu0303/VSAnet .

Generalization in image-based reinforcement learning (RL) aims to learn a robust policy that could be applied directly on unseen visual environments, which is a challenging task since agents usually tend to overfit to their training environment. To handle this problem, a natural approach is to increase the data diversity by image based augmentations. However, different with most vision tasks such as classification and detection, RL tasks are not always invariant to spatial based augmentations due to the entanglement of environment dynamics and visual appearance. In this paper, we argue with two principles for augmentations in RL: First, the augmented observations should facilitate learning a universal policy, which is robust to various distribution shifts. Second, the augmented data should be invariant to the learning signals such as action and reward. Following these rules, we revisit image-based RL tasks from the view of frequency domain and propose a novel augmentation method, namely Spectrum Random Masking (SRM),which is able to help agents to learn the whole frequency spectrum of observation for coping with various distributions and compatible with the pre-collected action and reward corresponding to original observation. Extensive experiments conducted on DMControl Generalization Benchmark demonstrate the proposed SRM achieves the state-of-the-art performance with strong generalization potentials.

Open set recognition (OSR), aiming to simultaneously classify the seen classes and identify the unseen classes as 'unknown', is essential for reliable machine learning. The key challenge of OSR is how to reduce the empirical classification risk on the labeled known data and the open space risk on the potential unknown data simultaneously. To handle the challenge, we formulate the open space risk problem from the perspective of multi-class integration, and model the unexploited extra-class space with a novel concept Reciprocal Point. Follow this, a novel learning framework, termed Adversarial Reciprocal Point Learning (ARPL), is proposed to minimize the overlap of known distribution and unknown distributions without loss of known classification accuracy. Specifically, each reciprocal point is learned by the extra-class space with the corresponding known category, and the confrontation among multiple known categories are employed to reduce the empirical classification risk. Then, an adversarial margin constraint is proposed to reduce the open space risk by limiting the latent open space constructed by reciprocal points. To further estimate the unknown distribution from open space, an instantiated adversarial enhancement method is designed to generate diverse and confusing training samples, based on the adversarial mechanism between the reciprocal points and known classes. This can effectively enhance the model distinguishability to the unknown classes. Extensive experimental results on various benchmark datasets indicate that the proposed method is significantly superior to other existing approaches and achieves state-of-the-art performance. The code is released on github.com/iCGY96/ARPL.

Online image hashing has received increasing research attention recently, which processes large-scale data in a streaming fashion to update the hash functions on-the-fly. To this end, most existing works exploit this problem under a supervised setting, i.e., using class labels to boost the hashing performance, which suffers from the defects in both adaptivity and efficiency: First, large amounts of training batches are required to learn up-to-date hash functions, which leads to poor online adaptivity. Second, the training is time-consuming, which contradicts with the core need of online learning. In this paper, a novel supervised online hashing scheme, termed Fast Class-wise Updating for Online Hashing (FCOH), is proposed to address the above two challenges by introducing a novel and efficient inner product operation. To achieve fast online adaptivity, a class-wise updating method is developed to decompose the binary code learning and alternatively renew the hash functions in a class-wise fashion, which well addresses the burden on large amounts of training batches. Quantitatively, such a decomposition further leads to at least 75 percent storage saving. To further achieve online efficiency, we propose a semi-relaxation optimization, which accelerates the online training by treating different binary constraints independently. Without additional constraints and variables, the time complexity is significantly reduced. Such a scheme is also quantitatively shown to well preserve past information during updating hashing functions. We have quantitatively demonstrated that the collective effort of class-wise updating and semi-relaxation optimization provides a superior performance comparing to various state-of-the-art methods, which is verified through extensive experiments on three widely-used datasets.

Neuronal circuits formed in the brain are complex with intricate connection patterns. Such complexity is also observed in the retina with a relatively simple neuronal circuit. A retinal ganglion cell (GC) receives excitatory inputs from neurons in previous layers as driving forces to fire spikes. Analytical methods are required to decipher these components in a systematic manner. Recently a method called spike-triggered non-negative matrix factorization (STNMF) has been proposed for this purpose. In this study, we extend the scope of the STNMF method. By using retinal GCs as a model system, we show that STNMF can detect various computational properties of upstream bipolar cells (BCs), including spatial receptive field, temporal filter, and transfer nonlinearity. In addition, we recover synaptic connection strengths from the weight matrix of STNMF. Furthermore, we show that STNMF can separate spikes of a GC into a few subsets of spikes, where each subset is contributed by one presynaptic BC. Taken together, these results corroborate that STNMF is a useful method for deciphering the structure of neuronal circuits.

Deep convolutional neural networks (CNNs) have demonstrated impressive performance on many visual tasks. Recently, they became useful models for the visual system in neuroscience. However, it is still not clear what is learned by CNNs in terms of neuronal circuits. When a deep CNN with many layers is used for the visual system, it is not easy to compare the structure components of CNNs with possible neuroscience underpinnings due to highly complex circuits from the retina to the higher visual cortex. Here, we address this issue by focusing on single retinal ganglion cells with biophysical models and recording data from animals. By training CNNs with white noise images to predict neuronal responses, we found that fine structures of the retinal receptive field can be revealed. Specifically, convolutional filters learned are resembling biological components of the retinal circuit. This suggests that a CNN learning from one single retinal cell reveals a minimal neural network carried out in this cell. Furthermore, when CNNs learned from different cells are transferred between cells, there is a diversity of transfer learning performance, which indicates that CNNs are cell specific. Moreover, when CNNs are transferred between different types of input images, here white noise versus natural images, transfer learning shows a good performance, which implies that CNNs indeed capture the full computational ability of a single retinal cell for different inputs. Taken together, these results suggest that CNNs could be used to reveal structure components of neuronal circuits, and provide a powerful model for neural system identification.

Event cameras, offering extremely high temporal resolution and high dynamic range, have brought a new perspective to addressing common object detection challenges (e.g., motion blur and low light). However, how to learn a better spatio-temporal representation and exploit rich temporal cues from asynchronous events for object detection still remains an open issue. To address this problem, we propose a novel asynchronous spatio-temporal memory network (ASTMNet) that directly consumes asynchronous events instead of event images prior to processing, which can well detect objects in a continuous manner. Technically, ASTMNet learns an asynchronous attention embedding from the continuous event stream by adopting an adaptive temporal sampling strategy and a temporal attention convolutional module. Besides, a spatio-temporal memory module is designed to exploit rich temporal cues via a lightweight yet efficient inter-weaved recurrent-convolutional architecture. Empirically, it shows that our approach outperforms the state-of-the-art methods using the feed-forward frame-based detectors on three datasets by a large margin (i.e., 7.6% in the KITTI Simulated Dataset, 10.8% in the Gen1 Automotive Dataset, and 10.5% in the 1Mpx Detection Dataset). The results demonstrate that event cameras can perform robust object detection even in cases where conventional cameras fail, e.g., fast motion and challenging light conditions.

Unsupervised domain adaptation (UDA) has achieved unprecedented success in improving the cross-domain robustness of object detection models. However, existing UDA methods largely ignore the instantaneous data distribution and the sampling strategy during model learning, which could deteriorate the feature representation given large domain shift. In this work, we propose a Self-Guided Adaptation (SGA) model, targeting at aligning feature representation and transferring object detection models across domains while considering the instantaneous alignment difficulty. The core of SGA is to calculate “hardness” factors for sample pairs indicating domain distance in a kernel space. With the hardness factor, the proposed SGA adaptively indicates the importance of samples and assigns them different constrains. Indicated by these hardness factors, Self-Guided Progressive Sampling (SPS) is implemented in an “easy-to-hard” way during model adaptation. Using multi-stage convolutional features, SGA is further aggregated to fully align hierarchical representations of detection models. Extensive experiments on commonly-used benchmarks show that SGA improves the state-of-the-art methods with significant margins especially on large domain shift cases.

We propose a novel network pruning approach by information preserving of pretrained network weights (filters). Network pruning with the information preserving is formulated as a matrix sketch problem, which is efficiently solved by the off-the-shelf frequent direction method. Our approach, referred to as FilterSketch, encodes the second-order information of pretrained weights, which enables the representation capacity of pruned networks to be recovered with a simple fine-tuning procedure. FilterSketch requires neither training from scratch nor data-driven iterative optimization, leading to a several-orders-of-magnitude reduction of time cost in the optimization of pruning. Experiments on CIFAR-10 show that FilterSketch reduces 63.3% of floating-point operations (FLOPs) and prunes 59.9% of network parameters with negligible accuracy cost for ResNet-110. On ILSVRC-2012, it reduces 45.5% of FLOPs and removes 43.0% of parameters with only 0.69% accuracy drop for ResNet-50. Our code and pruned models can be found at https://github.com/lmbxmu/FilterSketch.

Tracking-by-detection is a very popular framework for single-object tracking that attempts to search the target object within a local search window for each frame. Although such a local search mechanism works well on simple videos, however, it makes the trackers sensitive to extremely challenging scenarios, such as heavy occlusion and fast motion. In this article, we propose a novel and general target-aware attention mechanism (termed TANet) and integrate it with a tracking-by-detection framework to conduct joint local and global search for robust tracking. Specifically, we extract the features of the target object patch and continuous video frames; then, we concatenate and feed them into a decoder network to generate target-aware global attention maps. More importantly, we resort to adversarial training for better attention prediction. The appearance and motion discriminator networks are designed to ensure its consistency in spatial and temporal views. In the tracking procedure, we integrate target-aware attention with multiple trackers by exploring candidate search regions for robust tracking. Extensive experiments on both short- and long-term tracking benchmark datasets all validated the effectiveness of our algorithm.

Occluded person re-identification (re-ID) is a challenging task as different human parts may become invisible in cluttered scenes, making it hard to match person images of different identities. Most existing methods address this challenge by aligning spatial features of body parts according to semantic information (e.g. human poses) or feature similarities but this approach is complicated and sensitive to noises. This paper presents Matching on Sets (MoS), a novel method that positions occluded person re-ID as a set matching task without requiring spatial alignment. MoS encodes a person image by a pattern set as represented by a `global vector’ with each element capturing one specific visual pattern, and it introduces Jaccard distance as a metric to compute the distance between pattern sets and measure image similarity. To enable Jaccard distance over continuous real numbers, we employ minimization and maximization to approximate the operations of intersection and union, respectively. In addition, we design a Jaccard triplet loss that enhances the pattern discrimination and allows to embed set matching into deep neural networks for end-to-end training. In the inference stage, we introduce a conflict penalty mechanism that detects mutually exclusive patterns in the pattern union of image pairs and decreases their similarities accordingly. Extensive experiments over three widely used datasets (Market1501, DukeMTMC and Occluded-DukeMTMC) show that MoS achieves superior re-ID performance. Additionally, it is tolerant of occlusions and outperforms the state-of-the-art by large margins for Occluded-DukeMTMC.

Tracking by natural language specification is a new rising research topic that aims at locating the target object in the video sequence based on its language description. Compared with traditional bounding box (BBox) based tracking, this setting guides object tracking with high-level semantic information, addresses the ambiguity of BBox, and links local and global search organically together. Those benefits may bring more flexible, robust and accurate tracking performance in practical scenarios. However, existing natural language initialized trackers are developed and compared on benchmark datasets proposed for tracking-by-BBox, which can’t reflect the true power of tracking-by-language. In this work, we propose a new benchmark specifically dedicated to the tracking-by-language, including a large scale dataset, strong and diverse baseline methods. Specifically, we collect 2k video sequences (contains a total of 1,244,340 frames, 663 words) and split 1300/700 for the train/testing respectively. We densely annotate one sentence in English and corresponding bounding boxes of the target object for each video. We also introduce two new challenges into TNL2K for the object tracking task, i.e., adversarial samples and modality switch. A strong baseline method based on an adaptive local-global-search scheme is proposed for future works to compare. We believe this benchmark will greatly boost related researches on natural language guided tracking.

Fovea, located in the centre of the retina, is specialized for high-acuity vision. Mimicking the sampling mechanism of the fovea, a retina-inspired camera, named spiking camera, is developed to record the external information with a sampling rate of 40,000 Hz, and outputs asynchronous binary spike streams. Although the temporal resolution of visual information is improved, how to reconstruct the scenes is still a challenging problem. In this paper, we present a novel high-speed image reconstruction model through the short-term plasticity (STP) mechanism of the brain. We derive the relationship between postsynaptic potential regulated by STP and the firing frequency of each pixel. By setting up the STP model at each pixel of the spiking camera, we can infer the scene radiance with the temporal regularity of the spike stream. Moreover, we show that STP can be used to distinguish the static and motion areas and further enhance the reconstruction results. The experimental results show that our methods achieve state-of-the-art performance in both image quality and computing time.

Existing compression artifacts reduction methods aim to restore images on pixel-level, which can improve the human visual experience. However, in many applications, large-scale images are collected not for visual examination by humans. Instead, they are used for many high-level vision tasks usually by Deep Neural Networks (DNN). In this paper, we find that these methods have limited performance improvements to high-level tasks, even bring negative effects. Therefore, inspired by the teacher-student network framework, we propose a compression artifacts reduction framework (ARF) for DNN. In addition, we generalize our method to the unsupervised setting (U-ARF) where the corresponding original images are unavailable in training. Extensive experiments indicate the proposed methods can help DNNs improve performance on the highly compressed images significantly.

Short videos, ranging from a few seconds to a few minutes, have become popular in many video production. So we need efficient coding tools to compress these videos. AV1 is an emerging open-source and royalty-free video compression format jointly developed and finalized in early 2018 by the Alliance for Open Media. Many coding tools are introduced into AV1 to improve compression performance. However, some coding tools may lead to performance loss to encode these short videos. This paper presents a coding tools performance evaluation of AV1 on short videos. We design an Off experiment by turning off a single coding tool, and the anchor is AV1 with all coding tools on. We list the impact of each open and closed coding tool on the overall performance of the short video.

This paper presents an overview of the emerging area of collaborative intelligence (CI). Our goal is to raise awareness in the signal processing community of the challenges and opportunities in this area of growing importance, where key developments are expected to come from signal processing and related disciplines. The paper surveys the current state of the art in CI, with special emphasis on signal processing-related challenges in feature compression, error resilience, privacy, and system-level design.

With the development of intelligent hardware, front-end devices can also perform DNN computation. Moreover, the deep neural network can be divided into several layers. In this way, part of the computation of DNN models can be migrated to the front-end devices, which can alleviate the cloud burden and shorten the processing latency. This paper proposes a computation allocation algorithm of DNN between the front-end devices and the cloud server. In brief, we divide the DNN layers dynamically according to the current and the predicted future status of the processing system, by which we obtain a shorter end-to-end latency. The simulation results reveal that the overall latency reduction is more than 70% compared with traditional cloud-centered processing.

Neuromorphic vision sensor is a new bio-inspired imaging paradigm that emerged in recent years, which continuously sensing luminance intensity and firing asynchronous spikes (events) with high temporal resolution. Typically, there are two types of neuromorphic vision sensors, namely dynamic vision sensor (DVS) and spike camera. From the perspective of bio-inspired sampling, DVS only perceives movement by imitating the retinal periphery, while the spike camera was developed to perceive fine textures by simulating the fovea. It is meaningful to explore how to combine two types of neuromorphic cameras to reconstruct high quality image like human vision. In this paper, we propose a NeuSpike-Net to learn both the high dynamic range and high motion sensitivity of DVS and the full texture sampling of spike camera to achieve high-speed and high dynamic image reconstruction. We propose a novel representation to effectively extract the temporal information of spike and event data. By introducing the feature fusion module, the two types of neuromorphic data achieve complementary to each other. The experimental results on the simulated and real datasets demonstrate that the proposed approach is effective to reconstruct high-speed and high dynamic range images via the combination of spike and event data.

Recently, the generalization behavior of Convolutional Neural Networks (CNN) is gradually transparent through explanation techniques with the frequency components decomposition. However, the importance of the phase spectrum of the image for a robust vision system is still ignored. In this paper, we notice that the CNN tends to converge at the local optimum which is closely related to the high-frequency components of the training images, while the amplitude spectrum is easily disturbed such as noises or common corruptions. In contrast, more empirical studies found that humans rely on more phase components to achieve robust recognition. This observation leads to more explanations of the CNN’s generalization behaviors in both robustness to common perturbations and out-of-distribution detection, and motivates a new perspective on data augmentation designed by re-combing the phase spectrum of the current image and the amplitude spectrum of the distracter image. That is, the generated samples force the CNN to pay more attention to the structured information from phase components and keep robust to the variation of the amplitude. Experiments on several image datasets indicate that the proposed method achieves state-of-the-art performances on multiple generalizations and calibration tasks, including adaptability for common corruptions and surface variations, out-of-distribution detection, and adversarial attack. The code is released on github/iCGY96/APR.

Spiking Neural Networks (SNNs) have attracted enormous research interest due to temporal information processing capability, low power consumption, and high biological plausibility. However, the formulation of efficient and high-performance learning algorithms for SNNs is still challenging. Most existing learning methods learn weights only, and require manual tuning of the membrane-related parameters that determine the dynamics of a single spiking neuron. These parameters are typically chosen to be the same for all neurons, which limits the diversity of neurons and thus the expressiveness of the resulting SNNs. In this paper, we take inspiration from the observation that membrane-related parameters are different across brain regions, and propose a training algorithm that is capable of learning not only the synaptic weights but also the membrane time constants of SNNs. We show that incorporating learnable membrane time constants can make the network less sensitive to initial values and can speed up learning. In addition, we reevaluate the pooling methods in SNNs and find that max-pooling will not lead to significant information loss and have the advantage of low computation cost and binary compatibility. We evaluate the proposed method for image classification tasks on both traditional static MNIST, Fashion-MNIST, CIFAR-10 datasets, and neuromorphic N-MNIST, CIFAR10-DVS, DVS128 Gesture datasets. The experiment results show that the proposed method outperforms the state-of-the-art accuracy on nearly all datasets, using fewer time-steps. Our codes are available at https://github.com/fangwei123456/Parametric-Leaky-Integrate-and-Fire-Spiking-Neuron.

The crucial problem in vehicle re-identification is to find the same vehicle identity when reviewing this object from cross-view cameras, which sets a higher demand for learning viewpoint-invariant representations. In this paper, we propose to solve this problem from two aspects: constructing robust feature representations and proposing camera-sensitive evaluations. We first propose a novel Heterogeneous Relational Complement Network (HRCN) by incorporating region-specific features and cross-level features as complements for the original high-level output. Considering the distributional differences and semantic misalignment, we propose graph-based relation modules to embed these heterogeneous features into one unified high-dimensional space. On the other hand, considering the deficiencies of cross-camera evaluations in existing measures (i.e., CMC and AP), we then propose a Cross-camera Generalization Measure (CGM) to improve the evaluations by introducing position-sensitivity and cross-camera generalization penalties. We further construct a new benchmark of existing models with our proposed CGM and experimental results reveal that our proposed HRCN model achieves new state-of-the-art in VeRi-776, VehicleID, and VERI-Wild.

Spiking Neural Networks (SNNs) have been attached great importance due to their biological plausibility and high energy-efficiency on neuromorphic chips. As these chips are usually resource-constrained, the compression of SNNs is thus crucial along the road of practical use of SNNs. Most existing methods directly apply pruning approaches in artificial neural networks (ANNs) to SNNs, which ignore the difference between ANNs and SNNs, thus limiting the performance of the pruned SNNs. Besides, these methods are only suitable for shallow SNNs. In this paper, inspired by synaptogenesis and synapse elimination in the neural system, we propose gradient rewiring (Grad R), a joint learning algorithm of connectivity and weight for SNNs, that enables us to seamlessly optimize network structure without retraining. Our key innovation is to redefine the gradient to a new synaptic parameter, allowing better exploration of network structures by taking full advantage of the competition between pruning and regrowth of connections. The experimental results show that the proposed method achieves minimal loss of SNNs' performance on MNIST and CIFAR-10 datasets so far. Moreover, it reaches a ~3.5% accuracy loss under unprecedented 0.73% connectivity, which reveals remarkable structure refining capability in SNNs. Our work suggests that there exists extremely high redundancy in deep SNNs. Our codes are available at https://github.com/Yanqi-Chen/Gradient-Rewiring.

Spiking Neural Networks (SNNs), as bio-inspired energy-efficient neural networks, have attracted great attentions from researchers and industry. The most efficient way to train deep SNNs is through ANN-SNN conversion. However, the conversion usually suffers from accuracy loss and long inference time, which impede the practical application of SNN. In this paper, we theoretically analyze ANN-SNN conversion and derive sufficient conditions of the optimal conversion. To better correlate ANN-SNN and get greater accuracy, we propose Rate Norm Layer to replace the ReLU activation function in source ANN training, enabling direct conversion from a trained ANN to an SNN. Moreover, we propose an optimal fit curve to quantify the fit between the activation value of source ANN and the actual firing rate of target SNN. We show that the inference time can be reduced by optimizing the upper bound of the fit curve in the revised ANN to achieve fast inference. Our theory can explain the existing work on fast reasoning and get better results. The experimental results show that the proposed method achieves near loss-less conversion with VGG-16, PreActResNet-18, and deeper structures. Moreover, it can reach 8.6× faster reasoning performance under 0.265× energy consumption of the typical method. The code is available at https://github.com/DingJianhao/OptSNNConvertion-RNL-RIL.

The low-latency streams captured by event cameras have shown impressive potential in addressing vision tasks such as video reconstruction and optical flow estimation. However, these tasks often require massive training event streams, which are expensive to collect and largely bypassed by recently proposed event camera simulators. To align the statistics of synthetic events with that of target event cameras, existing simulators often need to be heuristically tuned with elaborative manual efforts and thus become incompetent to automatically adapt to various domains. To address this issue, this work proposes one of the first learning-based, domain-adaptive event simulator. Given a specific domain, the proposed simulator learns pixel-wise distributions of event contrast thresholds that, after stochastic sampling and paralleled rendering, can generate event representations well aligned with those from the data from realistic event cameras. To achieve such domain-specific alignment, we design a novel divide-and-conquer discrimination scheme that adaptively evaluates the synthetic-to-real consistency of event representations according to the local statistics of images and events. Trained with the data synthesized by the proposed simulator, the performances of state-of-the-art event-based video reconstruction and optical flow estimation approaches are boosted up to 22.9% and 2.8%, respectively. In addition, we show significantly improved domain adaptation capability over existing event simulators and tuning strategies, consistently on three real event datasets.

Conventional frame-based cameras for multimedia computing have encountered important challenges in high-speed and extreme light scenarios. However, how to design a novel paradigm for visual perception that overcomes the disadvantages of conventional cameras still remains an open issue. In this paper, we propose a novel solution, namely retinomorphic sensing, which integrates fovea-like and peripheral-like sampling mechanisms to generate asynchronous visual streams using a unified representation as the retina does. Technically, our encoder incorporates an interaction controller to switch flexibly between dynamic and static sensing. Then, the decoder effectively extracts dynamic events for machine vision and reconstructs visual textures for human vision. The results show that our strategy enables it to sense dynamic events and visual textures meanwhile reduce data redundancy. We further build a prototype hybrid camera system to verify this strategy on vision tasks such as image reconstruction and object detection. We believe that this novel paradigm will provide insight into future multimedia computing. The code can be available at https://github.com/acmmm2021-bni-retinomorphic/retinomorphic-sensing.

Deep learning has achieved great success in recognizing video actions, but the collection and annotation of training data are still quite laborious, which mainly lies in two aspects: (1) the amount of required annotated data is large; (2) temporally annotating the location of each action is time-consuming. Works such as few-shot learning or untrimmed video recognition have been proposed to handle either one aspect or the other. However, very few existing works can handle both issues simultaneously. In this paper, we target a new problem, Annotation-Efficient Video Recognition, to reduce the requirement of annotations for both large amount of samples and the action location. Such problem is challenging due to two aspects: (1) the untrimmed videos only have weak supervision; (2) video segments not relevant to current actions of interests (background, BG) could contain actions of interests (foreground, FG) in novel classes, which is a widely existing phenomenon but has rarely been studied in few-shot untrimmed video recognition. To achieve this goal, by analyzing the property of BG, we categorize BG into informative BG (IBG) and non-informative BG (NBG), and we propose (1) an open-set detection based method to find the NBG and FG, (2) a contrastive learning method to learn IBG and distinguish NBG in a self-supervised way, and (3) a self-weighting mechanism for the better distinguishing of IBG and FG. Extensive experiments on ActivityNet v1.2 and ActivityNet v1.3 verify the rationale and effectiveness of the proposed methods.

Existing few-shot learning (FSL) methods usually assume base classes and novel classes are from the same domain (in-domain setting). However, in practice, it may be infeasible to collect sufficient training samples for some special domains to construct base classes. To solve this problem, cross-domain FSL (CDFSL) is proposed very recently to transfer knowledge from general-domain base classes to special-domain novel classes. Existing CDFSL works mostly focus on transferring between near domains, while rarely consider transferring between distant domains, which is in practical need as any novel classes could appear in real-world applications, and is even more challenging. In this paper, we study a challenging subset of CDFSL where the novel classes are in distant domains from base classes, by revisiting the mid-level features, which are more transferable yet under-explored in main stream FSL work. To boost the discriminability of mid-level features, we propose a residual-prediction task to encourage mid-level features to learn discriminative information of each sample. Notably, such mechanism also benefits the in-domain FSL and CDFSL in near domains. Therefore, we provide two types of features for both cross- and in-domain FSL respectively, under the same training framework. Experiments under both settings on six public datasets, including two challenging medical datasets, validate the our rationale and demonstrate state-of-the-art performance. Code will be released.

Deep Spiking Neural Networks (SNNs) present optimization difficulties for gradient-based approaches due to discrete binary activation and complex spatial-temporal dynamics. Considering the huge success of ResNet in deep learning, it would be natural to train deep SNNs with residual learning. Previous Spiking ResNet mimics the standard residual block in ANNs and simply replaces ReLU activation layers with spiking neurons, which suffers the degradation problem and can hardly implement residual learning. In this paper, we propose the spike-element-wise (SEW) ResNet to realize residual learning in deep SNNs. We prove that the SEW ResNet can easily implement identity mapping and overcome the vanishing/exploding gradient problems of Spiking ResNet. We evaluate our SEW ResNet on ImageNet, DVS Gesture, and CIFAR10-DVS datasets, and show that SEW ResNet outperforms the state-of-the-art directly trained SNNs in both accuracy and time-steps. Moreover, SEW ResNet can achieve higher performance by simply adding more layers, providing a simple method to train deep SNNs. To our best knowledge, this is the first time that directly training deep SNNs with more than 100 layers becomes possible. Our codes are available at https://github.com/fangwei123456/Spike-Element-Wise-ResNet.

The human visual system can detect objects of unseen categories from merely a few examples. However, such capability remains absent in state-of-the-art detectors. To bridge this gap, several attempts have been proposed to perform few-shot detection by incorporating meta-learning techniques. Such methods can improve detection performance on unseen categories, but also add huge computational burden, and usually degrade detection performance on seen categories. In this paper, we present PNPDet, a novel Plug-and-Play Detector, for efficient few-shot detection without forgetting. It introduces a simple but effective architecture with separate sub-networks that disentangles the recognition of base and novel categories and prevents hurting performance on known categories while learning new concepts. Distance metric learning is further incorporated into sub-networks, consistently boosting detection performance for both base and novel categories. Experiments show that the proposed PNPDet can achieve comparable few-shot detection performance on unseen categories while not losing accuracy on seen categories, and also remain efficient and flexible at the same time.

Semantic object part segmentation is a fundamental task in object understanding and geometric analysis. The clear understanding of part relationships can be of great use to the segmentation process. In this work, we propose a novel Ordinal Multi-task Part Segmentation (OMPS) approach which explicitly models the part ordinal relationship to guide the segmentation process in a recurrent manner. Quantitative and qualitative experiments are conducted first to explore the mutual impacts among object parts and then an ordinal part inference algorithm is formulated via experimental observations. Specifically, our framework is mainly composed of two modules, the forward module to segment multiple parts as individual subtasks with prior knowledge, and the recurrent module to generate appropriate part priors with the ordinal inference algorithm. These two modules work iteratively to optimize the segmentation performance and the network parameters. Experimental results show that our approach outperforms the state-of-the-art models on human and vehicle part parsing benchmarks. Comprehensive evaluations are conducted to demonstrate the effectiveness of our approach in object part segmentation.

Neural architecture search (NAS) has achieved unprecedented performance in various computer vision tasks. However, most existing NAS methods are defected in search efficiency and model generalizability. In this paper, we propose a novel NAS framework, termed MIGO-NAS, with the aim to guarantee the efficiency and generalizability in arbitrary search spaces. On the one hand, we formulate the search space as a multivariate probabilistic distribution, which is then optimized by a novel multivariate information-geometric optimization (MIGO). By approximating the distribution with a sampling, training, and testing pipeline, MIGO guarantees the memory efficiency, training efficiency, and search flexibility. Besides, MIGO is the first time to decrease the estimation error of natural gradient in multivariate distribution. On the other hand, for a set of specific constraints, the neural architectures are generated by a novel dynamic programming network generation (DPNG), which significantly reduces the training cost under various hardware environments. Experiments validate the advantages of our approach over existing methods by establishing a superior accuracy and efficiency i.e., 2.39 test error on CIFAR-10 benchmark and 21.7 on ImageNet benchmark, with only 1.5 GPU hours and 96 GPU hours for searching, respectively. Besides, the searched architectures can be well generalize to computer vision tasks including object detection and semantic segmentation, i.e., 25× FLOPs compression, with 6.4 mAP gain over Pascal VOC dataset, and 29.9× FLOPs compression, with only 1.41 percent performance drop over Cityscapes dataset. The code is publicly available.

The ubiquitous camera networks in the city brain system grow at a rapid pace, creating massive amounts of images and videos at a range of spatial-temporal scales and thereby forming the “biggest” big data. However, the sensing system often lags behind the construction of the fast-growing city brain system, in the sense that such exponentially growing data far exceed today’s sensing capabilities. Therefore, critical issues arise regarding how to better leverage the existing city brain system and significantly improve the city-scale performance in intelligent applications. To tackle the unprecedented challenges, we articulate a vision towards a novel visual computing framework, termed asdigital retina, which aligns high-efficiency sensing models with the emerging Visual Coding for Machine (VCM) paradigm. In particular, digital retina may consist of video coding, feature coding, model coding, as well as their joint optimization. The digital retina is biologically-inspired, rooted on the widely accepted view that the retina encodes the visual information for human perception, and extracts features by the brain downstream areas to disentangle the visual objects. Within the digital retina framework, three streams, i.e., video stream, feature stream, and model stream, work collaboratively over the end-edge-cloud platform. In particular, the compressed video stream serves for human vision, the compact feature stream targets for machine vision, and the model stream incrementally updates deep learning models to improve the performance of human/machine vision tasks. We have developed a prototype to demonstrate the technical advantages of digital retina, and extensive experiments have been conducted to validate that it is able to effectively support the video big data analysis and retrieval in the intelligent city system. In particular, up to$7000\times $compression ratio could be realized for visual data compression while maintaining competitive performance with pristine signal in a series of visual analysis tasks.

Most of the existing single object trackers track the target in a unitary local search window, making them particularly vulnerable to challenging factors such as heavy occlusions and out-of-view movements. Despite the attempts to further incorporate global search, prevailing mechanisms that cooperate local and global search are relatively static, thus are still sub-optimal for improving tracking performance. By further studying the local and global search results, we raise a question: can we allow more dynamics for cooperating both results? In this paper, we propose to introduce more dynamics by devising a dynamic attention-guided multi-trajectory tracking strategy. In particular, we construct dynamic appearance model that contains multiple target templates, each of which provides its own attention for locating the target in the new frame. Guided by different attention, we maintain diversified tracking results for the target to build multi-trajectory tracking history, allowing more candidates to represent the true target trajectory. After spanning the whole sequence, we introduce a multi-trajectory selection network to find the best trajectory that deliver improved tracking performance. Extensive experimental results show that our proposed tracking strategy achieves compelling performance on various large-scale tracking benchmarks. The project page of this paper can be found athttps://sites.google.com/view/mt-track/.

Recently, a novel bio-inspired camera was developed by mimicking the retina fovea to continuously accumulate luminance intensity and then fire spikes once the dispatch threshold is reached. In contrast to the conventional frame-based cameras and the emerging dynamic vision sensors, this spike camera has shown remarkable advantages in capturing fast-moving scenes in a frame-free manner with full texture reconstruction capabilities. However, the ultra-high temporal resolution makes the transmission or storage of the output data of spike camera (referred to as spike data) quite difficult. To address the above challenges, we propose a unified lossy spike coding framework, which exploits the motion patterns hidden in the spike data distribution to design the motion-fidelity coding modes for the first time. We investigate the spatiotemporal distribution of spike data and propose an intensity-based measurement of the spike train distance. Then, the adaptive polyhedron partitioning is proposed to deal with the spike data with different motion characteristics. Finally, the intra-/inter-polyhedron prediction with spike-time and spike-rate modes, transform and multi-layer quantization are proposed and introduced into the codec. We also construct a PKU-Spike dataset captured by the spike camera to evaluate the compression performance. The experimental results on the dataset demonstrate that the proposed approach is effective in compressing such spike data while maintaining the visual fidelity especially for high-speed scenarios.

Hyperspectral image (HSI) restoration is to recover the clean image from degraded version, such as the noisy, blurred, or damaged. Recent low-rank tensor-based recovery methods have been widely explored in HSIs restoration. Most of previous methods, however, neglect an inconspicuous but important phenomenon that the physical meaning and dimension along the spatial, spectral, and nonlocal mode are markedly different. In this work, we discover the low-rank property discrepancy along spatial, spectral, and nonlocal self-similarity mode in the HSIs, and argue that the intrinsic low-rank correlations along each mode contribute different to the final restoration results. Consequently, we figure out that the combination of the spectral and nonlocal-induced low-rank is most beneficial for HSIs modeling, and propose an optimal low-rank tensor (OLRT) model for HSIs restoration. Furthermore, we not only explore the low-rank property in the image component, but also in the sparse error component (stripe noise in HSIs). Thus, we extend OLRT to the OLRT-robust principal component analysis (RPCA) with low-rank tensor priors for both the HSIs and sparse error. Besides, previous methods are usually designed for one specific HSI task, which is less robust to various tasks. We prove that the proposed optimal low-rank prior is very flexible for various HSI restoration problems including denoising, deblurring, inpainting, and destriping. The proposed methods have been extensively evaluated on several benchmarks and tasks, and greatly outperform state-of-the-art (STOA). We show the simple yet effective OLRT strategy is also beneficial to STOA.

Image-based salient object detection has made great progress over the past decades, especially after the revival of deep neural networks. By the aid of attention mechanisms to weight the image features adaptively, recent advanced deep learning-based models encourage the predicted results to approximate the ground-truth masks with as large predictable areas as possible, thus achieving the state-of-the-art performance. However, these methods do not pay enough attention to small areas prone to misprediction. In this way, it is still tough to accurately locate salient objects due to the existence of regions with indistinguishable foreground and background and regions with complex or fine structures. To address these problems, we propose a novel convolutional neural network with purificatory mechanism and structural similarity loss. Specifically, in order to better locate preliminary salient objects, we first introduce the promotion attention, which is based on spatial and channel attention mechanisms to promote attention to salient regions. Subsequently, for the purpose of restoring the indistinguishable regions that can be regarded as error-prone regions of one model, we propose the rectification attention, which is learned from the areas of wrong prediction and guide the network to focus on error-prone regions thus rectifying errors. Through these two attentions, we use the Purificatory Mechanism to impose strict weights with different regions of the whole salient objects and purify results from hard-to-distinguish regions, thus accurately predicting the locations and details of salient objects. In addition to paying different attention to these hard-to-distinguish regions, we also consider the structural constraints on complex regions and propose the Structural Similarity Loss. The proposed loss models the region-level pair-wise relationship between regions to assist these regions to calibrate their own saliency values. In experiments, the proposed purificatory mechanism and structural similarity loss can both effectively improve the performance, and the proposed approach outperforms 19 state-of-the-art methods on six datasets with a notable margin. Also, the proposed method is efficient and runs at over 27FPS on a single NVIDIA 1080Ti GPU.

Fine-grained visual recognition is to classify objects with visually similar appearances into subcategories, which has made great progress with the development of deep CNNs. However, handling subtle differences between different subcategories still remains a challenge. In this paper, we propose to solve this issue in one unified framework from two aspects, i.e., constructing feature-level interrelationships, and capturing part-level discriminative features. This framework, namely PArt-guided Relational Transformers (PART), is proposed to learn the discriminative part features with an automatic part discovery module, and to explore the intrinsic correlations with a feature transformation module by adapting the Transformer models from the field of natural language processing. The part discovery module efficiently discovers the discriminative regions which are highly-corresponded to the gradient descent procedure. Then the second feature transformation module builds correlations within the global embedding and multiple part embedding, enhancing spatial interactions among semantic pixels. Moreover, our proposed approach does not rely on additional part branches in the inference time and reaches state-of-the-art performance on 3 widely-used fine-grained object recognition benchmarks. Experimental results and explainable visualizations demonstrate the effectiveness of our proposed approach.

Foreground detection is one of the key steps in computer vision applications. Many foreground and background models have been proposed and achieved promising performance in static scenes. However, due to challenges such as dynamic background, irregular movement, and noise, most algorithms degrade sharply in complex scenes. To address the problem, we propose a motion-aware structured matrix factorization approach (MSMF), which integrates the structural and spatiotemporal motion information into a unified sparse-low-rank matrix factorization framework. Technologically, it has three main contributions: First, a variant of structured sparsity-inducing norm is proposed to constrain both structure and sparsity of foreground. The model is robust to the statistical variability of the underlying foreground pixels in complex scenes. Second, to capture the ambiguous pixels, a spatiotemporal cube-based motion trajectory is extracted for assisting matrix factorization. Finally, to solve the optimization problem of structured matrix factorization, we develop an augmented Lagrange multiplier method with the alternating direction strategy and Douglas-Rachford monotone operator splitting algorithm. Experiments demonstrate that the proposed approach achieves impressive performance in separating irregular moving foreground while suppressing the dynamic background and the noise, and outperforms some state-of-the-art algorithms.

Low bit-width model quantization is highly desirable when deploying a deep neural network on mobile and edge devices. Quantization is an effective way to reduce the model size with low bit-width weight representation. However, the unacceptable accuracy drop hinders the development of this approach. One possible reason for this is that the weights in quantization intervals are directly assigned to the center. At the same time, some quantization applications are limited by the various of different network models. Accordingly, in this paper, we propose Multiple Phase Adaptations (MPA), a framework designed to address these two problems. Firstly, weights in the target interval are assigned to center by gradually spreading the quantization range. During the MPA process, the accuracy drop can be compensated for the unquantized parts. Moreover, as MPA does not introduce hyperparameters that depend on different models or bit-width, the framework can be conveniently applied to various models. Extensive experiments demonstrate that MPA achieves higher accuracy than most existing methods on classification tasks for AlexNet, VGG-16 and ResNet.

Neural network pruning offers a promising prospect to facilitate deploying deep neural networks on resource-limited devices. However, existing methods are still challenged by the training inefficiency and labor cost in pruning designs, due to missing theoretical guidance of non-salient network components. In this paper, we propose a novel filter pruning method by exploring the High Rank of feature maps (HRank). Our HRank is inspired by the discovery that the average rank of multiple feature maps generated by a single filter is always the same, regardless of the number of image batches CNNs receive. Based on HRank, we develop a method that is mathematically formulated to prune filters with low-rank feature maps. The principle behind our pruning is that low-rank feature maps contain less information, and thus pruned results can be easily reproduced. Besides, we experimentally show that weights with high-rank feature maps contain more important information, such that even when a portion is not updated, very little damage would be done to the model performance. Without introducing any additional constraints, HRank leads to significant improvements over the state-of-the-arts in terms of FLOPs and parameters reduction, with similar accuracies. For example, with ResNet-110, we achieve a 58.2%-FLOPs reduction by removing 59.2% of the parameters, with only a small loss of 0.14% in top-1 accuracy on CIFAR-10. With Res-50, we achieve a 43.8%-FLOPs reduction by removing 36.7% of the parameters, with only a loss of 1.17% in the top-1 accuracy on ImageNet. The codes can be available at https://github.com/lmbxmu/HRank.

Domain adaptive person re-identification (re-ID) is a challenging task, especially when person identities in target domains are unknown. Existing methods attempt to address this challenge by transferring image styles or aligning feature distributions across domains, whereas the rich unlabeled samples in target domains are not sufficiently exploited. This paper presents a novel augmented discriminative clustering (AD-Cluster) technique that estimates and augments person clusters in target domains and enforces the discrimination ability of re-ID models with the augmented clusters. AD-Cluster is trained by iterative density-based clustering, adaptive sample augmentation, and discriminative feature learning. It learns an image generator and a feature encoder which aim to maximize the intra-cluster diversity in the sample space and minimize the intra-cluster distance in the feature space in an adversarial min-max manner. Finally, AD-Cluster increases the diversity of sample clusters and improves the discrimination capability of re-ID models greatly. Extensive experiments over Market-1501 and DukeMTMC-reID show that AD-Cluster outperforms the state-of-the-art with large margins.

Neural architecture search (NAS) remains a challenging problem, which is attributed to the indispensable and time-consuming component of performance estimation (PE). In this paper, we provide a novel yet systematic rethinking of PE in a resource constrained regime, termed budgeted PE (BPE), which precisely and effectively estimates the performance of an architecture sampled from an architecture space. Since searching an optimal BPE is extremely time-consuming as it requires to train a large number of networks for evaluation, we propose a Minimum Importance Pruning (MIP) approach. Given a dataset and a BPE search space, MIP estimates the importance of hyper-parameters using random forest and subsequently prunes the minimum one from the next iteration. In this way, MIP effectively prunes less important hyper-parameters to allocate more computational resource on more important ones, thus achieving an effective exploration. By combining BPE with various search algorithms including reinforcement learning, evolution algorithm, random search, and differentiable architecture search, we achieve 1, 000× of NAS speed up with a negligible performance drop comparing to the SOTA.

The high-sensitivity vision of primates, including humans, is mediated by a small retinal region called the fovea. As a novel bio-inspired vision sensor, spike camera mimics the fovea to record the nature scenes by continuous-time spikes instead of frame-based manner. However, reconstructing visual images from the spikes remains to be a challenge. In this paper, we design a retina-like visual image reconstruction framework, which is flexible in reconstructing full texture of natural scenes from the totally new spike data. Specifically, the proposed architecture consists of motion local excitation layer, spike refining layer and visual reconstruction layer motivated by bio-realistic leaky integrate and fire (LIF) neurons and synapse connection with spike-timing-dependent plasticity (STDP) rules. This approach may represent a major shift from conventional frame-based vision to the continuous-time retina-like vision, owning to the advantages of high temporal resolution and low power consumption. To test the performance, a spike dataset is constructed which is recorded by the spike camera. The experimental results show that the proposed approach is extremely effective in reconstructing the visual image in both normal and high speed scenes, while achieving high dynamic range and high image quality.

A common framework of the action recognition is to collect the videos from different cameras into a cloud center firstly, and then perform the 3D CNN on the cloud server. Although directly, this framework will bring a huge burden to the cloud server and video transmission. To handle this challenge, the "front-cloud" collaborative processing architecture can be used. The most import issue is to compress the feature from 3D CNN effectively without significant loss of accuracy. We propose logarithmic quantization with a maximum value threshold and HEVC inter encoding for 3D CNN features. Experimental results on ResNet-50 and InceptionV1 show that the features can be represented by only 1 bit without significant loss of accuracy. The compression ratio of the quantized 1 bit features using HEVC inter coding can reach to 5000 times and the loss of accuracy is less than 1%.

Channel pruning is among the predominant approaches to compress deep neural networks. To this end, most existing pruning methods focus on selecting channels (filters) by importance/optimization or regularization based on rule-of-thumb designs, which defects in sub-optimal pruning. In this paper, we propose a new channel pruning method based on artificial bee colony algorithm (ABC), dubbed as ABCPruner, which aims to efficiently find optimal pruned structure, i.e., channel number in each layer, rather than selecting "important" channels as previous works did. To solve the intractably huge combinations of pruned structure for deep networks, we first propose to shrink the combinations where the preserved channels are limited to a specific space, thus the combinations of pruned structure can be significantly reduced. And then, we formulate the search of optimal pruned structure as an optimization problem and integrate the ABC algorithm to solve it in an automatic manner to lessen human interference. ABCPruner has been demonstrated to be more effective, which also enables the fine-tuning to be conducted efficiently in an end-to-end manner. The source codes can be available at https: //github.com/lmbxmu/ABCPruner.

Owing to the rich emerging multimedia applications and services in the past decade, super large amount of multimedia data has been produced for the purpose of advanced research in multimedia. Furthermore, multimedia research has made great progress on image/video content analysis, multimedia search and recommendation, multimedia streaming, multimedia content delivery etc. At the same time, Artificial Intelligence (AI) has undergone a "new" wave of development since being officially regarded as an academic discipline in 1950s, which should give credits to the extreme success of deep learning. Thus, one question naturally arises: What happens when multimedia meets Artificial Intelligence?

This paper focuses on a novel task named masked faces recognition (MFR), which aims to match masked faces with common faces and is important especially during the global outbreak of COVID-19. It is challenging to identify masked faces for two main reasons. Firstly, there is no large-scale training data and test data with ground truth for MFR. Collecting and annotating millions of masked faces is labor-consuming. Secondly, since most facial cues are occluded by mask, it is necessary to learn representations which are both discriminative and robust to mask wearing. To handle the first challenge, this paper collects two datasets designed for MFR: MFV with 400 pairs of 200 identities for verification, and MFI which contains 4,916 images of 669 identities for identification. As is known, a robust face recognition model needs images of millions of identities to train, and hundreds of identities is far from enough. Hence, MFV and MFI are only considered as test datasets to evaluate algorithms. Besides, a data augmentation method for training data is introduced to automatically generate synthetic masked face images from existing common face datasets. In addition, a novel latent part detection (LPD) model is proposed to locate the latent facial part which is robust to mask wearing, and the latent part is further used to extract discriminative features. The proposed LPD model is trained in an end-to-end manner and only utilizes the original and synthetic training data. Experimental results on MFV, MFI and synthetic masked LFW demonstrate that LPD model generalizes well on both realistic and synthetic masked data and outperforms other methods by a large margin.

Masked faces recognition (MFR) aims to match a masked face with its corresponding full face, which is an important task especially during the global outbreak of COVID-19. However, most existing face recognition models generalize poorly in this case, and it is hard to train a robust MFR model due to two main reasons: 1) the absence of large scale training data as well as ground truth testing data, and 2) the presence of large intra-class variation between masked faces and full faces. To address the first challenge, this paper firstly contributes a new dataset denoted as MFSR, which consists of two parts. The first part contains 9,742 masked face images with mask region segmentation annotation. The second part contains 11,615 images of 1,004 identities, and each identity has masked and full face images with various orientations, lighting conditions and mask types. However, it is still not enough for training MFR models with deep learning. To obtain sufficient training data, based on the MFSR, we introduce a novel Identity Aware Mask GAN (IAMGAN) with segmentation guided multi-level identity preserve module to generate the synthetic masked face images from the full face images. In addition, to tackle the second challenge, a Domain Constrained Ranking (DCR) loss is proposed by adopting a center-based cross-domain ranking strategy. For each identity, two centers are designed which correspond to the full face images and the masked face images respectively. The DCR forces the feature of masked faces getting closer to its corresponding full face center and vice-versa. Experimental results on the MFSR dataset demonstrate the effectiveness of the proposed approaches.

Person re-identification (ReID) aims to match detected pedestrian images from multiple non-overlapping cameras. Most existing methods employ a backbone CNN to extract a vectorized feature representation by performing some global pooling operations (such as global average pooling and global max pooling) on the 3D feature map (i.e., the output of the backbone CNN). Although simple and effective in some situations, the global pooling operation only focuses on the statistical properties and ignores the spatial distribution of the feature map. Hence, it can not distinguish two feature maps when they have similar response values located in totally different positions. To handle this challenge, a novel method is proposed to learn the discriminative spatial features. Firstly, a self-constrained spatial transformer network (SC-STN) is introduced to handle the misalignments caused by detection errors. Then, based on the prior knowledge that the spatial structure of a pedestrian often keeps robust in vertical orientation of images, a novel vertical convolution network (VCN) is proposed to extract the spatial feature in vertical. Extensive experimental evaluations on several benchmarks demonstrate that the proposed method achieves state-of-the-art performances by introducing only a few parameters to the backbone.

Given base classes with sufficient labeled samples, the target of few-shot classification is to recognize unlabeled samples of novel classes with only a few labeled samples. Most existing methods only pay attention to the relationship between labeled and unlabeled samples of novel classes, which do not make full use of information within base classes. In this paper, we make two contributions to investigate the few-shot classification problem. First, we report a simple and effective baseline trained on base classes in the way of traditional supervised learning, which can achieve comparable results to the state of the art. Second, based on the baseline, we propose a cooperative bi-path metric for classification, which leverages the correlations between base classes and novel classes to further improve the accuracy. Experiments on two widely used benchmarks show that our method is a simple and effective framework, and a new state of the art is established in the few-shot classification field.

Few-shot learning (FSL) aims at recognizing novel classes given only few training samples, which still remains a great challenge for deep learning. However, humans can easily recognize novel classes with only few samples. A key component of such ability is the compositional recognition that human can perform, which has been well studied in cognitive science but is not well explored in FSL. Inspired by such capability of humans, to imitate humans' ability of learning visual primitives and composing primitives to recognize novel classes, we propose an approach to FSL to learn a feature representation composed of important primitives, which is jointly trained with two parts, i.e. primitive discovery and primitive enhancing. In primitive discovery, we focus on learning primitives related to object parts by self-supervision from the order of image splits, avoiding extra laborious annotations and alleviating the effect of semantic gaps. In primitive enhancing, inspired by current studies on the interpretability of deep networks, we provide our composition view for the FSL baseline model. To modify this model for effective composition, inspired by both mathematical deduction and biological studies (the Hebbian Learning rule and the Winner-Take-All mechanism), we propose a soft composition mechanism by enlarging the activation of important primitives while reducing that of others, so as to enhance the influence of important primitives and better utilize these primitives to compose novel classes. Extensive experiments on public benchmarks are conducted on both the few-shot image classification and video recognition tasks. Our method achieves the state-of-the-art performance on all these datasets and shows better interpretability.

Skeleton-based action recognition has attracted more and more attention in recent years. Besides, the rapid development of deep learning has greatly improved the performance. However, the current exploration of action co-occurrence is still not comprehensive enough. Most existing works only mine co-occurrence features from the temporal or spatial domain seperately, and it's common to combine them in the end. Different from previous works, our approach is able to learn temporal and spatial co-occurrence features integratedly and globally, which is called spatio-temporal-unit feature enhancement (STUFE). In order to better align the skeleton data, we introduce a novel method for skeleton data preprocessing called active coordinate system conversion (ACSC). A coordinate system can be learned automatically to transform skeleton samples for alignment. By the way, the proposed methods are compatible with current two types of mainstream models, the CNN-based and GCN-based models. Finally, on the two benchmarks of NTU-RGB+D and SBU Kinect Interaction, we validated our methods based on two mainstream models. The results show that our methods achieve the state-of-the-art.

Practically, it is more feasible to collect compact visual features rather than the video streams from hundreds of thousands of cameras into the cloud for big data analysis and retrieval. Then the problem becomes which kinds of features should be extracted, compressed and transmitted so as to meet the requirements of various visual tasks. Recently, many studies have indicated that the activations from the convolutional layers in convolutional neural networks (CNNs) can be treated as local deep features describing particular details inside an image region, which are then aggregated (e.g., using Fisher Vectors) as a powerful global descriptor. Combination of local and global features can satisfy those various needs effectively. It has also been validated that, if only local deep features are coded and transmitted to the cloud while the global features are recovered using the decoded local features, the aggregated global features should be lossy and consequently would degrade the overall performance. Therefore, this paper proposes a joint coding framework for local and global deep features (DFJC) extracted from videos. In this framework, we introduce a coding scheme for real-valued local and global deep features with intra-frame lossy coding and inter-frame reference coding. The theoretical analysis is performed to understand how the number of inliers varies with the number of local features. Moreover, the inter-feature correlations are exploited in our framework. That is, local feature coding can be accelerated by making use of the frame types determined with global features, while the lossy global features aggregated with the decoded local features can be used as a reference for global feature coding. Extensive experimental results under three metrics show that our DFJC framework can significantly reduce the bitrate of local and global deep features from videos while maintaining the retrieval performance.

As an emerging vision platform, a drone can look from many abnormal viewpoints which brings many new challenges into the classic vision task of video saliency prediction. To investigate these challenges, this paper proposes a large-scale video dataset for aerial saliency prediction, which consists of ground-truth salient object regions of 1,000 aerial videos, annotated by 24 subjects. To the best of our knowledge, it is the first large-scale video dataset that focuses on visual saliency prediction on drones. Based on this dataset, we propose a Model-guided Multi-path Network (MM-Net) that serves as a baseline model for aerial video saliency prediction. Inspired by the annotation process in eye-tracking experiments, MM-Net adopts multiple information paths, each of which is initialized under the guidance of a classic saliency model. After that, the visual saliency knowledge encoded in the most representative paths is selected and aggregated to improve the capability of MM-Net in predicting spatial saliency in aerial scenarios. Finally, these spatial predictions are adaptively combined with the temporal saliency predictions via a spatiotemporal optimization algorithm. Experimental results show that MM-Net outperforms ten state-of-the-art models in predicting aerial video saliency.

The papers in this special section is to broadly engage the machine learning and multimedia communities on the emerging yet challenging interpretable machine learning. Multimedia is increasingly becoming the “biggest big data,” among the most important and valuable source for insight and information. Many powerful machine learning algorithms, especially deep learning models such as convolutional neural networks (CNNs), have recently achieved outstanding predictive performance in a wide range of multimedia applications, including visual object classification, scene understanding, speech recognition, and activity prediction. Nevertheless, most deep learning algorithms are generally conceived as blackbox methods, and it is difficult to intuitively and quantitatively understand the results of their prediction and inference. Since this lack of interpretability is a major bottleneck in designing more successful predictive models and exploring wider-range useful applications, there has been an explosion of interest in interpreting the representations learned by these models, with profound implications for research into interpretable machine learning in the multimedia community.

One-shot action recognition aims at recognizing actions in unseen classes in cases where only one training video is provided. Compared with one-shot image recognition, one-shot learning on videos is more difficult due to the fact that the temporal dimension of video may lead to greater variation. To handle this variation, it is important to conduct further adaptation in the one-shot training process, despite the scarcity of the training data. While meta-learning is an option for facilitating this adaptation, it cannot be directly applied for two reasons: first, deep networks for action recognition can make current meta-learning methods infeasible to run because of their high computational complexity; second, due to the greater variation in actions, the adapted performance may not be higher than the un-adapted one, making it difficult to train the model by means of meta-learning. To address these problems and facilitate the adaptation, we propose the Adaptation-Oriented Feature (AOF) projection for one-shot action recognition. We first pre-train the base network on seen classes. The output of the network is projected to the adaptation-oriented feature space by fusing the important feature dimensions that are sensitive to adaptation. Subsequently, a small dataset (a.k.a. task) is sampled from seen classes to simulate the unseen-class training and testing settings. The feature adaptation is performed on the training data of this task to integrate the distribution information of the adapted feature. In order to reduce over-fitting, the triplet loss is applied to handle temporal variation with fewer parameters during the adaptation. On the testing data of this task, the losses on both adapted and un-adapted features are calculated to train the projection matrix. This sampling-adaptation-training procedure is then repeated on seen classes until convergence. Extensive experimental results on two challenging one-shot action recognition datasets demonstrate that our proposed method outperforms state-of-the-art methods.

Vehicle re-identification (Re-ID) has been attracting more interests in computer vision owing to its great contributions in urban surveillance and intelligent transportation. With the development of deep learning approaches, vehicle Re-ID still faces a near-duplicate challenge, which is to distinguish different instances with nearly identical appearances. Previous methods simply rely on the global visual features to handle this problem. In this paper, we proposed a simple but efficient part-regularized discriminative feature preserving method which enhances the perceptive ability of subtle discrepancies. We further develop a novel framework to integrate part constrains with the global Re-ID modules by introducing an detection branch. Our framework is trained end-to-end with combined local and global constrains. Specially, without the part-regularized local constrains in inference step, our Re-ID network outperforms the state-of-the-art method by a large margin on large benchmark datasets VehicleID and VeRi-776.

Recently, a novel bio-inspired spike camera has been proposed, which continuously accumulates luminance intensity and fires spikes once the dispatch threshold is reached. It has shown great advantages in capturing fast-moving scene in a frame-free manner with full texture reconstruction capabilities. However, it is difficult to transmit or store the large amount of spike data. By investigating the spatiotemporal distribution of the spikes, we propose an intensity-based measurement for spike train distance and design an efficient coding method to meet the challenge. First, the spike train is transformed into inter-spike intervals (ISIs), and ISIs are adaptively partitioned into multiple segments in temporal. Then, intra-and inter-pixel prediction are performed to find the best reference candidate. The prediction residuals are quantized to achieve lossy compression. Finally, the quantized residuals are fed into an adaptive context-based entropy coder. Overall, to achieve the best performance, each prediction mode will be tried and the one with minimum rate-distortion cost is chosen.

Dynamic vision sensor (DVS) as a bio-inspired camera, has shown great advantages in high dynamic range (HDR) and high temporal resolution (us) in vision tasks. However, how to lossy compress asynchronous spikes for meeting the demand of large-scale transmission and storage meanwhile maintaining the analysis performance still remains open. Towards this end, this paper proposes a lossy spike coding framework for DVS.

In the modern video coding standards, transform coefficient coding takes large portion of output bit-stream. The absolute value of coefficients are usually represented with high-efficiency variable-length code, while the signs are binarized with 1-bit fixed-length code, which ignores the apriority constraints on sign combinations beyond signs. In this paper, we present a fast and efficient sign inference algorithm for improving the coding efficiency of coefficients signs. By considering the correlation of pixels across the transform block boundary, we define a criterion of continuity to constrain the combination space of signs in the block, and shorten the bits will be used. A given number of signs are hidden during bypass coding, and inferred with the high-precision criterion. Then the infer-error is coded with arithmetic coding. Meanwhile a sign bit redefining strategy is introduced, which is based on the invariants of the linear orthogonal separable transform. The strategy ensures that the codec is able to choose up to 4 bits for inferring each block, and avoids the complex decision mechanism introduced by rate distortion optimization. Experiment results show that about 0.6% in average and up to 1.1% BD-rate saving is archived and only less than 1% encoding time complexity increased compared with HEVC baseline.

Few-shot learning, which aims at extracting new concepts rapidly from extremely few examples of novel classes, has been featured into the meta-learning paradigm recently. Yet, the key challenge of how to learn a generalizable classifier with the capability of adapting to specific tasks with severely limited data still remains in this domain. To this end, we propose a Transductive Episodic-wise Adaptive Metric (TEAM) framework for few-shot learning, by integrating the meta-learning paradigm with both deep metric learning and transductive inference. With exploring the pairwise constraints and regularization prior within each task, we explicitly formulate the adaptation procedure into a standard semi-definite programming problem. By solving the problem with its closed-form solution on the fly with the setup of transduction, our approach efficiently tailors an episodic-wise metric for each task to adapt all features from a shared task-agnostic embedding space into a more discriminative task-specific metric space. Moreover, we further leverage an attention-based bi-directional similarity strategy for extracting the more robust relationship between queries and prototypes. Extensive experiments on three benchmark datasets show that our framework is superior to other existing approaches and achieves the state-of-the-art performance in the few-shot literature.

Typically, a salient object detection (SOD) model faces opposite requirements in processing object interiors and boundaries. The features of interiors should be invariant to strong appearance change so as to pop-out the salient object as a whole, while the features of boundaries should be selective to slight appearance change to distinguish salient objects and background. To address this selectivity-invariance dilemma, we propose a novel boundary-aware network with successive dilation for image-based SOD. In this network, the feature selectivity at boundaries is enhanced by incorporating a boundary localization stream, while the feature invariance at interiors is guaranteed with a complex interior perception stream. Moreover, a transition compensation stream is adopted to amend the probable failures in transitional regions between interiors and boundaries. In particular, an integrated successive dilation module is proposed to enhance the feature invariance at interiors and transitional regions. Extensive experiments on six datasets show that the proposed approach outperforms 16 state-of-the-art methods.

Object part parsing in the wild, which requires to simultaneously detect multiple object classes in the scene and accurately segments semantic parts within each class, is challenging for the joint presence of class-level and part-level ambiguities. Despite its importance, however, this problem is not sufficiently explored in existing works. In this paper, we propose a joint parsing framework with boundary and semantic awareness to address this challenging problem. To handle part-level ambiguity, a boundary awareness module is proposed to make mid-level features at multiple scales attend to part boundaries for accurate part localization, which are then fused with high-level features for effective part recognition. For class-level ambiguity, we further present a semantic awareness module that selects discriminative part features relevant to a category to prevent irrelevant features being merged together. The proposed modules are lightweight and implementation friendly, improving the performance substantially when plugged into various baseline architectures. Without bells and whistles, the full model sets new state-of-the-art results on the Pascal-Part dataset, in both multi-class and the conventional single-class setting, while running substantially faster than recent high-performance approaches.

Due to the high-speed motion blur and low dynamic range, conventional frame-based cameras have encountered an important challenge in object detection, especially in autonomous driving. Event-based cameras, by taking the advantages of high temporal resolution and high dynamic range, have brought a new perspective to address the challenge. Motivated by this fact, this paper proposes a joint framework combining event-based and frame-based vision for vehicle detection. Specially, two separate event-based and frame-based streams are incorporated into a convolutional neural network (CNN). Besides, to accommodate the asynchronous events from event-based cameras, a convolutional spiking neural network (SNN) is utilized to generate visual attention maps so that two streams can be synchronized. Moreover, Dempster-Shafer theory is introduced to merge two outputs from CNN in a joint decision model. The experimental results show that the proposed approach outperforms the state-of-the-art methods only using frame-based information, especially in fast motion and challenging illumination conditions.

Conventional frame-based camera is not able to meet the demand of rapid reaction for real-time applications, while the emerging dynamic vision sensor (DVS) can realize high speed capturing for moving objects. However, to achieve visual texture reconstruction, DVS need extra information apart from the output spikes. This paper introduces a fovea-like sampling method inspired by the neuron signal processing in retina, which aims at visual texture reconstruction only taking advantage of the properties of spikes. In the proposed method, the pixels independently respond to the luminance changes with temporal asynchronous spikes. Analyzing the arrivals of spikes makes it possible to restore the luminance information, enabling reconstructing the natural scene for visualization. Three decoding methods of spike stream for texture reconstruction are proposed for high-speed motion and stationary scenes. Compared to conventional frame-based camera and DVS, our model can achieve better image quality and higher flexibility, which is capable of changing the way that demanding machine vision applications are built.

It is a significant problem to predict the 2D LiDAR map at next moment for robotics navigation and path-planning. To tackle this problem, we resort to the motion flow between adjacent maps, as motion flow is a powerful tool to process and analyze the dynamic data, which is named optical flow in video processing. However, unlike video, which contains abundant visual features in each frame, a 2D LiDAR map lacks distinctive local features. To alleviate this challenge, we propose to estimate the motion flow based on deep neural networks inspired by its powerful representation learning ability in estimating the optical flow of the video. To this end, we design a recurrent neural network based on gated recurrent unit, which is named LiDAR-FlowNet. As a recurrent neural network can encode the temporal dynamic information, our LiDAR-FlowNet can estimate motion flow between the current map and the unknown next map only from the current frame and previous frames. A self-supervised strategy is further designed to train the LiDAR-FlowNet model effectively, while no training data need to be manually annotated. With the estimated motion flow, it is straightforward to predict the 2D LiDAR map at the next moment. Experimental results verify the effectiveness of our LiDAR-FlowNet as well as the proposed training strategy. The results of the predicted LiDAR map also show the advantages of our motion flow based method.

In recent years, camera-based localization has been widely used for robotic applications, and most proposed algorithms rely on local features extracted from recorded images. For better performance, the features used for open-loop localization are required to be short-term globally static, and the ones used for re-localization or loop closure detection need to be long-term static. Therefore, the motion attribute of a local feature point could be exploited to improve localization performance, e.g., the feature points extracted from moving persons or vehicles can be excluded from these systems due to their unsteadiness. In this paper, we design a fully convolutional network (FCN), named MD-Net, to perform motion attribute estimation and feature description simultaneously. MD-Net has a shared backbone network to extract features from the input image and two network branches to complete each sub-task. With MD-Net, we can obtain the motion attribute while avoiding increasing much more computation. Experimental results demonstrate that the proposed method can learn distinct local feature descriptor along with motion attribute only using an FCN, by outperforming competing methods by a wide margin. We also show that the proposed algorithm can be integrated into a vision-based localization algorithm to improve estimation accuracy significantly.

The intractable “cold-start” problem often encountered in existed video recommendation systems when a new video is coming with minimal users' feedback since most existing recommendation algorithms heavily rely on the users' implicit feedbacks. This paper presents a novel idea for solving the “cold start” problem by analyzing the video content itself. The Harmonic Sampling is designed in our work for utilizing the rank information automatically during the sampling procedure, and a Semi-Siamese network is proposed for overcoming the asymmetric training samples. The proposed method demonstrated its effectiveness in dealing with “cold-start” problem, achieving superior performance over the Content-based Video Relevance Prediction Dataset.

Skeleton-based action recognition continues to open up new application scenarios with the popularity of acquisition devices. This also leads to a rapid increase in the amount of human skeleton data. Currently, there is no skeleton data compression algorithm for the task of action recognition. In order to solve this problem, we propose the first skeleton data compression algorithm, which can compress the skeleton data stream to a small bandwidth while keeping the accuracy of action recognition as high as possible. The proposed compression algorithm is called Motion-based Joints Selection (MJS). It performs compression based on the amount of movement of different joints. In addition, we also explored the combination of MJS and existing lossless compression methods, and found the most suitable one. In the end, we verify that our compression method MJS can achieve promising results on the large dataset NTU-RGB+D.

Dynamic vision sensor (DVS) as a bio-inspired camera, has shown great advantages in wide dynamic range and high temporal resolution imaging in contrast to conventional frame-based cameras. Its ability to capture high speed moving objects enables fast and accurate detection which plays a significant role in the emerging intelligent driving applications. The pixels in DVS independently respond to the luminance changes with output spikes. Thus, the spike stream conveying the x -, y -addresses, the firing time, and the polarity (ON/OFF), is quite different from conventional video frames. How to compress this kind of new data for efficient transmission and storage remains a big challenge, especially for on-board detection, monitoring and recording. To address this challenge, this paper first analyzes the spike firing mechanism and the spatiotemporal characteristics of the spike data, then introduces a cube-based spike coding framework for DVS. In the framework, an octree-based structure is proposed to adaptively partition the spike stream into coding cubes in both spatial and temporal dimensions, then several prediction modes are designed to exploit the spatial and temporal characteristics of spikes for compression, including address-prior mode and time-prior mode. To explore more flexibility, the intercube prediction is discussed extensively involving motion estimation and motion compensation. Finally, the experimental results demonstrate that our approach achieves an impressive coding performance with the average compression ratio of 2.6536 against the raw spike data, which is much higher than the results of conventional lossless coding algorithms.

As an emerging kind of retinomorphic camera, the dynamic vision sensors (DVS) have shown great advantages in wide dynamic range and high temporal resolution in various applications such as autonomous driving and high-speed motion photography. However, how to compress the output spike data of DVS still remains a big challenge. To address this challenge, this paper firstly analyzes the spike firing mechanism and the redundancies of the spike data generated from DVS, and then introduces an efficient cube-based coding framework. Typically, a spike in DVS contains the location (the x-, y- addresses, the timestamp) and the polarity (On/Off). Three key strategies are designed to exploit the spatial and temporal characteristics of the spike location information for compression, including the adaptive macro-cube partitioning structure, the address-prior mode and the time-prior mode. Finally, the experimental results demonstrate that our approach achieves an impressive coding performance, with the average compression ratio of 19.519 over the original spike data, which is much higher than the results of conventional lossless coding algorithms.

In action recognition, one of the most important challenges is to jointly utilize the texture and motion information as well as capturing the long-term dependence of various common and action-specific postures. Motivated by this fact, this paper proposes Temporal Attentive Network (TAN) for action recognition. The key idea in TAN is that not all postures, each of which represented by a small collection of consecutive frames, contribute equally to the successful recognition of an action. As a result, TAN incorporates two separate spatial and temporal streams into one network. Information in the two streams is partially shared so that discriminative spatiotemporal features can be extracted to characterize various postures in an action. Moreover, a temporal attention mechanism is introduced in the form of Long-Short Term Memory (LSTM) network. With this mechanism, features from the action-specific postures can be emphasized, while common postures shared by many different actions will be ignored to some extent. By jointly using such spatial and temporal information as well as attentive cues in a single network, TAN achieves impressive performance on two public datasets, HMDB51 and UCF101, with accuracy scores of 72.5% and 94.1 %, respectively.

In recent years, the performance of action recognition has been significantly improved with the help of deep neural networks. Most of the existing action recognition works hold the closed-set assumption that all action categories are known beforehand while deep networks can be well trained for these categories. However, action recognition in the real world is essentially an open-set problem, namely, it is impossible to know all action categories beforehand and consequently infeasible to prepare sufficient training samples for those emerging categories. In this case, applying closed-set recognition methods will definitely lead to unseen-category errors. To address this challenge, we propose the Open Deep Network (ODN) for the open-set action recognition task. Technologically, ODN detects new categories by applying a multiclass triplet thresholding method, and then dynamically reconstructs the classification layer and “opens” the deep network by adding predictors for new categories continually. In order to transfer the learned knowledge to the new category, two novel methods, Emphasis Initialization and Allometry Training, are adopted to initialize and incrementally train the new predictor so that only few samples are needed to fine-tune the model. Extensive experiments show that ODN can effectively detect and recognize new categories with little human intervention, thus applicable to the open-set action recognition tasks in the real world. Moreover, ODN can even achieve comparable performance to some closed-set methods.

The weakly supervised object localization (WSOL) is to locate the objects in an image while only image-level labels are available during the training procedure. In this work, the Selective Feature Category Mapping (SFCM) method is proposed, which introduces the Feature Category Mapping (FCM) and the widely-used selective search method to solve the WSOL task. Our FCM replaces layers after the specific layer in the state-of-the-art CNNs with a set of kernels and learns the weighted pooling for previous feature maps. It is trained with only image-level labels and then map the feature maps to their corresponding categories in the test phase. Together with selective search method, the location of each object is finally obtained. Extensive experimental evaluation on ILSVRC2012 and PASCAL VOC2007 benchmarks shows that SFCM is simple but very effective, and it is able to achieve outstanding classification performance and outperform the state-of-the-art methods in the WSOL task.

One-shot action recognition is one of the most challenging tasks due to the very limited training samples. For one-shot video action recognition, randomly selected frames from cluttered frame features may result in a poor performance. To use the most valuable frames in a better feature space, this paper proposes Hierarchical Temporal Memory Enhanced One-shot Distance Learning (HED). Firstly, we introduce temporal triplet from different frames, so that the intra-class distance will be decreased while the inter-class distance will be increased. Secondly, the Hierarchical Temporal Memory (HTM), a biological plausible unsupervised model for sequence prediction, is employed to enhance the one-shot action recognition by finding the most valuable frames in a video sequence. Finally, the selected frames together with the temporal triplet trained model are used to get the corresponding category label. Extensive experiments conducted on three benchmark datasets (i.e UCF11, UCF50 and HMDB51) demonstrate that we can achieve significant improvement than the state-of-the-art methods.

Under person re-identification (Re-ID), a query photo of the target person is often required for retrieval. However, one is not always guaranteed to have such a photo readily available under a practical forensic setting. In this paper, we define the problem of Sketch Re-ID, which instead of using a photo as input, it initiates the query process using a professional sketch of the target person. This is akin to the traditional problem of forensic facial sketch recognition, yet with the major difference that our sketches are whole-body other than just the face. This problem is challenging because sketches and photos are in two distinct domains. Specifically, a sketch is the abstract description of a person. Besides, person appearance in photos is variational due to camera viewpoint, human pose and occlusion. We address the Sketch Re-ID problem by proposing a cross-domain adversarial feature learning approach to jointly learn the identity features and domain-invariant features. We employ adversarial feature learning to filter low-level interfering features and remain high-level semantic information. We also contribute to the community the first Sketch Re-ID dataset with 200 persons, where each person has one sketch and two photos from different cameras associated. Extensive experiments have been performed on the proposed dataset and other common sketch datasets including CUFSF and QUML-shoe. Results show that the proposed method outperforms the state-of-the-arts.

To convert the input into binary code, hashing algorithm has been widely used for approximate nearest neighbor search on large-scale image sets due to its computation and storage efficiency. Deep hashing further improves the retrieval quality by combining the hash coding with deep neural network. However, a major difficulty in deep hashing lies in the discrete constraints imposed on the network output, which generally makes the optimization NP hard. In this work, we adopt the greedy principle to tackle this NP hard problem by iteratively updating the network toward the probable optimal discrete solution in each iteration. A hash coding layer is designed to implement our approach which strictly uses the sign function in forward propagation to maintain the discrete constraints, while in back propagation the gradients are transmitted intactly to the front layer to avoid the vanishing gradients. In addition to the theoretical derivation, we provide a new perspective to visualize and understand the effectiveness and efficiency of our algorithm. Experiments on benchmark datasets show that our scheme outperforms state-of-the-art hashing methods in both supervised and unsupervised tasks.

A number of vision problems such as zero-shot learning and person re-identification can be considered as cross-class transfer learning problems. As mid-level semantic properties shared cross different object classes, attributes have been studied extensively for knowledge transfer across classes. Most previous attribute learning methods focus only on human-defined/nameable semantic attributes, whilst ignoring the fact there also exist undefined/latent shareable visual properties, or latent attributes. These latent attributes can be either discriminative or non-discriminative parts depending on whether they can contribute to an object recognition task. In this work, we argue that learning the latent attributes jointly with user-defined semantic attributes not only leads to better representation but also helps semantic attribute prediction. A novel dictionary learning model is proposed which decomposes the dictionary space into three parts corresponding to semantic, latent discriminative and latent background attributes respectively. Such a joint attribute learning model is then extended by following a multi-task transfer learning framework to address a more challenging unsupervised domain adaptation problem, where annotations are only available on an auxiliary dataset and the target dataset is completely unlabelled. Extensive experiments show that the proposed models, though being linear and thus extremely efficient to compute, produce state-of-the-art results on both zero-shot learning and person re-identification.

Summary form only given. This paper introduces a spike camera with a distinct video capture scheme and proposes two methods of decoding the spike stream for texture reconstruction. The spike camera captures light and accumulates the converted luminance intensity at each pixel. A spike is fired when the accumulated intensity exceeds the dispatch threshold. The spike stream generated by the camera indicates the luminance variation. Analyzing the patterns of the spike stream makes it possible to reconstruct the picture of any moment which enables the playback of high speed movement.

Despite a lot of research efforts devoted in recent years, how to efficiently learn long-term dependencies from sequences still remains a pretty challenging task. As one of the key models for sequence learning, recurrent neural network (RNN) and its variants such as long short term memory (LSTM) and gated recurrent unit (GRU) are still not powerful enough in practice. One possible reason is that they have only feedforward connections, which is different from the biological neural system that is typically composed of both feedforward and feedback connections. To address this problem, this paper proposes a biologicallyinspired deep network, called shuttleNet. Technologically, the shuttleNet consists of several processors, each of which is a GRU while associated with multiple groups of hidden states. Unlike traditional RNNs, all processors inside shuttleNet are loop connected to mimic the brain's feedforward and feedback connections, in which they are shared across multiple pathways in the loop connection. Attention mechanism is then employed to select the best information flow pathway. Extensive experiments conducted on two benchmark datasets (i.e UCF101 and HMDB51) show that we can beat state-of-the-art methods by simply embedding shuttleNet into a CNN-RNN framework.

Precise search of visually-similar vehicles poses a great challenge in computer vision, which needs to find exactly the same vehicle among a massive vehicles with visually similar appearances for a given query image. In this paper, we model the relationship of vehicle images as multiple grains. Following this, we propose two approaches to alleviate the precise vehicle search problem by exploiting multi-grain ranking constraints. One is Generalized Pairwise Ranking, which generalizes the conventional pairwise from considering only binary similar/dissimilar relations to multiple relations. The other is Multi-Grain based List Ranking, which introduces permutation probability to score a permutation of a multi-grain list, and further optimizes the ranking by the likelihood loss function. We implement the two approaches with multi-attribute classification in a multi-task deep learning framework. To further facilitate the research on precise vehicle search, we also contribute two high-quality and well-annotated vehicle datasets, named VD1 and VD2, which are collected from two different cities with diverse annotated attributes. As two of the largest publicly available precise vehicle search datasets, they contain 1,097,649 and 807,260 vehicle images respectively. Experimental results show that our approaches achieve the state-of-the-art performance on both datasets.

In this paper, we propose a new approach for searching action proposals in unconstrained videos. Our method first produces snippet action proposals by combining state-of-the-art YOLO detector (Static YOLO) and our regression based RNN detector (Recurrent YOLO). Then, these short action proposals are integrated to form final action proposals by solving two-pass dynamic programming which maximizes actioness score and temporal smoothness concurrently. Our experimental comparison with other state-of-the-arts on challenging UCF101 dataset shows that our method advances state-of-the-art proposal generation performance while maintaining low computational cost.

Perception is the interaction interface between an intelligent system and the real world. Without sophisticated and flexible perceptual capabilities, it is impossible to create advanced artificial intelligence (AI) systems. For the next-generation AI, called ‘AI 2.0’, one of the most significant features will be that AI is empowered with intelligent perceptual capabilities, which can simulate human brain’s mechanisms and are likely to surpass human brain in terms of performance. In this paper, we briefly review the state-of-the-art advances across different areas of perception, including visual perception, auditory perception, speech perception, and perceptual information processing and learning engines. On this basis, we envision several R&D trends in intelligent perception for the forthcoming era of AI 2.0, including: (1) human-like and transhuman active vision; (2) auditory perception and computation in an actual auditory setting; (3) speech perception and computation in a natural interaction setting; (4) autonomous learning of perceptual information; (5) large-scale perceptual information processing and learning platforms; and (6) urban omnidirectional intelligent perception and reasoning engines. We believe these research directions should be highlighted in the future plans for AI 2.0.

With the explosion in the use of cameras in mobile phones or video surveillance systems, it is impossible to transmit a large amount of videos captured from a wide area into a cloud for big data analysis and retrieval. Instead, a feasible solution is to extract and compress features from videos and then transmit the compact features to the cloud. Meanwhile, many recent studies also indicate that the features extracted from the deep convolutional neural networks will lead to high performance for various analysis and recognition tasks. However, how to compress video deep features meanwhile maintaining the analysis or retrieval performance still remains open. To address this problem, we propose a high-efficiency deep feature coding (DFC) framework in this paper. In the DFC framework, we define three types of features in a group-of-features (GOFs) according to their coding modes (i.e., I-feature, P-feature, and S-feature). We then design two prediction structures for these features in a GOF, including a sequential prediction structure and an adaptive prediction structure. Similar to video coding, it is important for P-feature residual coding optimization to make a tradeoff between feature bitrate and analysis/retrieval performance when encoding residuals. To do so, we propose a rate-performance-loss optimization model. To evaluate various feature coding methods for large-scale video retrieval, we construct a video feature coding data set, called VFC-1M, which consists of uncompressed videos from different scenarios captured from real-world surveillance cameras, with totally 1M visual objects. Extensive experiments show that the proposed DFC can significantly reduce the bitrate of deep features in the videos while maintaining the retrieval accuracy.

Learning the spatial-temporal representation of motion information is crucial to human action recognition. Nevertheless, most of the existing features or descriptors cannot capture motion information effectively, especially for long-term motion. To address this problem, this paper proposes a long-term motion descriptor called sequential deep trajectory descriptor (sDTD). Specifically, we project dense trajectories into two-dimensional planes, and subsequently a CNN-RNN network is employed to learn an effective representation for long-term motion. Unlike the popular two-stream ConvNets, the sDTD stream is introduced into a three-stream framework so as to identify actions from a video sequence. Consequently, this three-stream framework can simultaneously capture static spatial features, short-term motion, and long-term motion in the video. Extensive experiments were conducted on three challenging datasets: KTH, HMDB51, and UCF101. Experimental results show that our method achieves state-of-the-art performance on the KTH and UCF101 datasets, and is comparable to the state-of-the-art methods on the HMDB51 dataset.

In visual saliency estimation, one of the most challenging tasks is to distinguish targets and distractors that share certain visual attributes. With the observation that such targets and distractors can sometimes be easily separated when projected to specific subspaces, we propose to estimate image saliency by learning a set of discriminative subspaces that perform the best in popping out targets and suppressing distractors. Toward this end, we first conduct principal component analysis on massive randomly selected image patches. The principal components, which correspond to the largest eigenvalues, are selected to construct candidate subspaces since they often demonstrate impressive abilities to separate targets and distractors. By projecting images onto various subspaces, we further characterize each image patch by its contrasts against randomly selected neighboring and peripheral regions. In this manner, the probable targets often have the highest responses, while the responses at background regions become very low. Based on such random contrasts, an optimization framework with pairwise binary terms is adopted to learn the saliency model that best separates salient targets and distractors by optimally integrating the cues from various subspaces. Experimental results on two public benchmarks show that the proposed approach outperforms 16 state-of-the-art methods in human fixation prediction.

The growing explosion in the use of surveillance cameras in public security highlights the importance of vehicle search from a large-scale image or video database. However, compared with person re-identification or face recognition, vehicle search problem has long been neglected by researchers in vision community. This paper focuses on an interesting but challenging problem, vehicle re-identification (a.k.a precise vehicle search). We propose a Deep Relative Distance Learning (DRDL) method which exploits a two-branch deep convolutional network to project raw vehicle images into an Euclidean space where distance can be directly used to measure the similarity of arbitrary two vehicles. To further facilitate the future research on this problem, we also present a carefully-organized largescale image database "VehicleID", which includes multiple images of the same vehicle captured by different realworld cameras in a city. We evaluate our DRDL method on our VehicleID dataset and another recently-released vehicle model classification dataset "CompCars" in three sets of experiments: vehicle re-identification, vehicle model verification and vehicle retrieval. Experimental results show that our method can achieve promising results and outperforms several state-of-the-art approaches.

Most existing person re-identification (Re-ID) approaches follow a supervised learning framework, in which a large number of labelled matching pairs are required for training. This severely limits their scalability in realworld applications. To overcome this limitation, we develop a novel cross-dataset transfer learning approach to learn a discriminative representation. It is unsupervised in the sense that the target dataset is completely unlabelled. Specifically, we present an multi-task dictionary learning method which is able to learn a dataset-shared but target-data-biased representation. Experimental results on five benchmark datasets demonstrate that the method significantly outperforms the state-of-the-art.

In the past decade, SIFT is widely used in most vision tasks such as image retrieval. While in recent several years, deep convolutional neural networks (CNN) features achieve the state-of-the-art performance in several tasks such as image classification and object detection. Thus a natural question arises: for the image retrieval task, can CNN features substitute for SIFT? In this paper, we experimentally demonstrate that the two kinds of features are highly complementary. Following this fact, we propose an image representation model, complementary CNN and SIFT (CCS), to fuse CNN and SIFT in a multi-level and complementary way. In particular, it can be used to simultaneously describe scene-level, object-level and point-level contents in images. Extensive experiments are conducted on four image retrieval benchmarks, and the experimental results show that our CCS achieves state-of-the-art retrieval results.

Human gait has been shown to be an efficient biometric measure for person identification at a distance. However, it often needs different gait features to handle various covariate conditions including viewing angles, walking speed, carrying an object and wearing different types of shoes. In order to improve the robustness of gait-based person re-identification on such multi-covariate conditions, a novel Swiss-system based cascade ranking model is proposed in this paper. Since the ranking model is able to learn a subspace where the potential true match is given the highest ranking, we formulate the gait-based person re-identification as a bipartite ranking problem and utilize it as an effective way for multi-feature ensemble learning. Then a Swiss multi-round competition system is developed for the cascade ranking model to optimize its effectiveness and efficiency. Extensive experiments on three indoor and outdoor public datasets demonstrate that our model outperforms several state-of-the-art methods remarkably.

In this paper, we propose a robust method to remove motion blur from a single photograph. We find that an inaccurate kernel and an unreliable final latent image reconstruction method are two main factors leading to low-quality restored images. To improve image quality, we do the following technical contributions. For robust blur kernel estimation, first, an edge mask and a smooth constraint are used to provide reliable intermediate latent images for salient structure extraction; second, we adopt an effective salient structure selection method to remove detrimental edges for kernel estimation; third, we use a gradient sparsity prior to remove kernel noise and ensure the continuity of blur kernels. For final latent image reconstruction, we combine the merits of both the TV-l2model and the hyper-Laplacian model to preserve tiny details and eliminate noise. Experimental results on synthetically blurred images and real photographs demonstrate that the proposed algorithm performs better than state-of-the-art approaches.

In this paper, we present a novel framework to detect abnormal behaviors in surveillance videos by using fuzzy clustering and multiple Auto-Encoders (FMAE). As detecting abnormal behaviors is often treated as an unsupervised task, how to describe normal patterns becomes the key point. Considering there are many types of normal behaviors in the daily life, we use the fuzzy clustering technique to roughly divide the training samples into several clusters so that each cluster stands for a normal pattern. Then we deploy multiple Auto-Encoders to estimate these different types of normal behaviors from weighted samples. When testing on an unknown video, our framework can predict whether it contains abnormal behaviors or not by summarizing the reconstruction cost through each Auto-Encoder. Since there are always lots of redundancies in the surveillance video, Auto-Encoder is a pretty good tool to capture common structures of normal video sequences automatically as well as estimate normal patterns. The experimental results show that our approach achieves good performance on three public video analysis datasets and statistically outperforms the state-of-the-art approaches under some scenes.

A remarkable compress performance was achieved in the surveillance video coding when high-quality reconstructed background frames were referenced in long-term. However, the high-quality reconstructed background frames lead to bursting bit-rate peaks in the video transmission, which may cause obvious communication delay or buffer overflow. In order to address this problem, the paper proposes a quality-progressive coding algorithm for smoothing the bursting peaks caused by high-quality reconstructed background frames. Instead of a single high-quality reconstructed background frame, we divide the background frame into a set of frames, which include the basic reconstructed background frame of normal-quality (basic part) and a series of reconstructed residual frames (residual part), for transmission. Moreover, the modeled background frame and the residual frames should be encoded into the bit-stream and transmitted every several frames, and the coding bits of two frame types above should be limited to the target range of coding bits, which is based on the channel capacity, so that avoids the bursting bit-rate peak and the transmission delay. Background frames are reconstructed by summing up the basic part and reconstructed residual frames one by one, and the last reconstructed background frame becomes a high-quality reconstructed background frame. Of course, each reconstructed background frame act as a prediction reference for its subsequent frames. Experimental results on an opening dataset, PKU-SVD-A, show that the proposed approach can smooth the bit-rate of high-quality reconstructed background frame in surveillance video coding, and achieve 0.57% bit-rate saving on average compared with HEVC-S.

There are two sides to every story of visual saliency modeling in the frequency domain. On the one hand, image saliency can be effectively estimated by applying simple operations to the frequency spectrum. On the other hand, it is still unclear which part of the frequency spectrum contributes the most to popping-out targets and suppressing distractors. Toward this end, this paper tentatively explores the secret of image saliency in the frequency domain. From the results obtained in several qualitative and quantitative experiments, we find that the secret of visual saliency may mainly hide in the phases of intermediate frequencies. To explain this finding, we reinterpret the concept of discrete Fourier transform from the perspective of template-based contrast computation and thus develop several principles for designing the saliency detector in the frequency domain. Following these principles, we propose a novel approach to design the saliency detector under the assistance of prior knowledge obtained through both unsupervised and supervised learning processes. Experimental results on a public image benchmark show that the learned saliency detector outperforms 18 state-of-the-art approaches in predicting human fixations.

The goal of this special issue is to provide a premier forum for researchers to present their recent research results on multimedia big data. It follows the recent success event—the First IEEE International Conference on Multimedia Big Data (BigMM 2015) that took place at the Chinese National Convention Center in Beijing, China, from April 20–22, 2015 It also provides an important opportunity for multidisciplinary work connecting big data to multimedia computing.

How to precisely and efficiently detect near-duplicate copies with complicated audiovisual transformations from a large-scale video database is a challenging task. To cope with this challenge, this article proposes a transformation-aware soft cascading (TASC) approach for multimodal video copy detection. Basically, our approach divides query videos into some categories and then for each category designs a transformation-aware chain to organize several detectors in a cascade structure. In each chain, efficient but simple detectors are placed in the forepart, whereas effective but complex detectors are located in the rear. To judge whether two videos are near-duplicates, a Detection-on-Copy-Units mechanism is introduced in the TASC, which makes the decision of copy detection depending on the similarity between their most similar fractions, called copy units (CUs), rather than the video-level similarity. Following this, we propose a CU search algorithm to find a pair of CUs from two videos and a CU-based localization algorithm to find the precise locations of their copy segments that are with the asserted CUs as the center. Moreover, to address the problem that the copies and noncopies are possibly linearly inseparable in the feature space, the TASC also introduces a flexible strategy, called soft decision boundary , to replace the single threshold strategy for each detector. Its basic idea is to automatically learn two thresholds for each detector to examine the easy-to-judge copies and noncopies, respectively, and meanwhile to train a nonlinear classifier to further check those hard-to-judge ones. Extensive experiments on three benchmark datasets showed that the TASC can achieve excellent copy detection accuracy and localization precision with a very high processing efficiency.

As the image enhancement algorithms developed in recent years, how to compare the performances of different image enhancement algorithms becomes a novel task. In this paper, we propose a framework to do quality assessment for comparing image enhancement algorithms. Not like traditional image quality assessment approaches, we focus on the relative quality ranking between enhanced images rather than giving an absolute quality score for a single enhanced image. We construct a dataset which contains source images in bad visibility and their enhanced images processed by different enhancement algorithms, and then do subjective assessment in a pair-wise way to get the relative ranking of these enhanced images. A rank function is trained to fit the subjective assessment results, and can be used to predict ranks of new enhanced images which indicate the relative quality of enhancement algorithms. The experimental results show that our proposed approach statistically outperforms state-of-the-art general-purpose NR-IQA algorithms.

Changes in the viewing angles pose a major challenge for gait recognition because the human gait silhouettes can be different under the various viewing angles. Recently, View Transformation Model (VTM) was proposed to tackle this problem by transforming gait features from across views to a common viewing angle. However, VTM must use the data of subjects crossing all views to train the pre-constructed model, which might be unsuitable for the real applications. To address this problem, this paper proposes a View Feature Recovering Model (VFRM) to generate the VTM with incomplete training data. In our algorithm, if the gait signature of a pedestrian is missing under a view, it can be recovered from the K-nearest pedestrians whose gait features are available in the same view. Moreover, the Geodesic distance based K-Nearest Neighbor (GKNN) algorithm is adopted in our algorithm to better measure the neighborhood between two pedestrians. Experimental results on a benchmark database has demonstrated the effectiveness of our method.

Basically, motion search is very time-consuming in the process of video coding. For surveillance videos, however, there exist a large amount of static background regions whose motion vectors actually are equal to zero. By utilizing the background and foreground information of coding units, this paper proposes a background-foreground division based search algorithm (BFDS) to accelerate the motion search in surveillance video coding. The basic idea of BFDS is to classify a predicting unit (PU) into a background predicting unit (BPU) or a foreground predicting unit (FPU) and then adopt different search strategies respectively for BPUs and FPUs. That is, a zero motion vector biased search strategy is applied in BPUs to reduce the search complexity on a large scale while a precise global search strategy is applied in FPUs to get higher coding performance. Compared with the current TZ search algorithm used in HEVC, the proposed BFDS algorithm can reduce the number of search points by 57.73% while remaining the coding performance almost unchanged.

Object detection plays a fundamental role in many content-based video systems. Often, it is still challenging to achieve both a reasonable accuracy and a fairly fast processing speed. In this paper, we propose a new object detection framework which utilizes raw RGB data from the pixel domain and some useful coding information from the compressed domain jointly. Firstly, various pixel-level detection algorithms can be embedded in our framework so as to obtain the preliminary results. Then by segmenting the moving regions from the background with the Hit-times Map (HTM), some false results can be removed and meanwhile the detection process can also be accelerated since the search area for sliding the detection window has been restricted to relatively small regions. After that, an additional regulation process is performed to further refine the preliminary detection results by employing both temporal consistency and spatial compactness in the motion vector(MV) field. The experimental results on two benchmark datasets show that the proposed method achieves a remarkable improvement both in detection accuracy and processing speed.

The exponential growth of surveillance videos presents an unprecedented challenge for high-efficiency surveillance video coding technology. Compared with the existing coding standards that were basically developed for generic videos, surveillance video coding should be designed to make the best use of the special characteristics of surveillance videos (e.g., relative static background). To do so, this paper first conducts two analyses on how to improve the background and foreground prediction efficiencies in surveillance video coding. Following the analysis results, we propose a background-modeling-based adaptive prediction (BMAP) method. In this method, all blocks to be encoded are firstly classified into three categories. Then, according to the category of each block, two novel inter predictions are selectively utilized, namely, the background reference prediction (BRP) that uses the background modeled from the original input frames as the long-term reference and the background difference prediction (BDP) that predicts the current data in the background difference domain. For background blocks, the BRP can effectively improve the prediction efficiency using the higher quality background as the reference; whereas for foreground-background-hybrid blocks, the BDP can provide a better reference after subtracting its background pixels. Experimental results show that the BMAP can achieve at least twice the compression ratio on surveillance videos as AVC (MPEG-4 Advanced Video Coding) high profile, yet with a slightly additional encoding complexity. Moreover, for the foreground coding performance, which is crucial to the subjective quality of moving objects in surveillance videos, BMAP also obtains remarkable gains over several state-of-the-art methods.

For the real-time and low-delay video surveillance and teleconferencing applications, the newly video coding standard HEVC can achieve much higher coding efficiency over H.264/AVC. However, we still argue that the hierarchical prediction structure in the HEVC low-delay encoder still does not fully utilize the special characteristics of surveillance and conference videos that are usually captured by stationary cameras. In this case, the background picture (G-picture), which is modeled from the original input frames, can be used to further improve the HEVC low-delay coding efficiency meanwhile reducing the complexity. Therefore, we propose an optimization method for the hierarchical prediction and coding in HEVC for these videos with background modeling. First, several experimental and theoretical analyses are conducted on how to utilize the G-picture to optimize the hierarchical prediction structure and hierarchical quantization. Following these results, we propose to encode the G-picture as the long-term reference frame to improve the background prediction, and then present a G-picture-based bit-allocation algorithm to increase the coding efficiency. Meanwhile, according to the proportions of background and foreground pixels in coding units (CUs), an adaptive speed-up algorithm is developed to classify each CU into different categories and then adopt different speed-up strategies to reduce the encoding complexity. To evaluate the performance, extensive experiments are performed on the HEVC test model. Results show our method can averagely save 39.09% bits and reduce the encoding complexity by 43.63% on surveillance videos, whereas those are 5.27% and 43.68% on conference videos.

Cloud is becoming a dominant computing platform. Naturally, a question that arises is whether we can beat notorious DDoS attacks in a cloud environment. Researchers have demonstrated that the essential issue of DDoS attack and defense is resource competition between defenders and attackers. A cloud usually possesses profound resources and has full control and dynamic allocation capability of its resources. Therefore, cloud offers us the potential to overcome DDoS attacks. However, individual cloud hosted servers are still vulnerable to DDoS attacks if they still run in the traditional way. In this paper, we propose a dynamic resource allocation strategy to counter DDoS attacks against individual cloud customers. When a DDoS attack occurs, we employ the idle resources of the cloud to clone sufficient intrusion prevention servers for the victim in order to quickly filter out attack packets and guarantee the quality of the service for benign users simultaneously. We establish a mathematical model to approximate the needs of our resource investment based on queueing theory. Through careful system analysis and real-world data set experiments, we conclude that we can defeat DDoS attacks in a cloud environment.

Summary form only given. Recently, a low-delay and high-efficiency hierarchical prediction structure (HPS) has been proposed for the forthcoming HEVC. Actually, frames and coding units (CUs) at different HPS positions have different importance to predict following frames and CUs. This paper firstly analyzes what frames and CUs should be quantified less. Based on the analysis, we propose a Hierarchical-and-Adaptive BIT-allocation method with Selective background prediction (HABITS) to optimize the video performance of HEVC. Extensive experiments on HM8.0 show that, HABITS saves 13.3% and 35.5% of the total bit rate for eight HEVC conference videos and eight common used surveillance videos. Even for the normal videos in HEVC's Class B and C, there is still 2.2% bit-saving.

Event detection in crowded surveillance videos is a challenging yet important problem. In this paper, we present our eSur (Event detection system on SURveillance video) system, which is derived from TRECVid'12 surveillance tasks. Currently, eSur attempts to detect two categories of events: 1) pair-wise events (e.g., PeopleMeet, PeopleSplitUp and Embrace); 2) action-like events (e.g., ObjectPut, CellToEar, PersonRuns and Pointing). In eSur system, we first employ people detection and tracking algorithms to locate target persons in 3D space-time domain. Then the video sequences in which target persons occur are partitioned into several spatio-temporal cubes. Visual features (i.e. cubic feature and MoSIFT) are computed over these cubes. After that, a sequence learning method, (namely SVM with dynamic time alignment kernel), is employed to infer the existence of an event for the video sequence. According to the TRECVid SED formal evaluation, eSur has yielded fairly encouraging results on TRECVid'12 dataset.

Among the multiple application-oriented video groups of IEEE 1857 video part, surveillance groups are the first specific video coding standards targeting on the exploring surveillance system. In this paper, we firstly present an overview of the technical features and characteristics of the Surveillance Baseline and Surveillance Groups. The video coding technologies are then described in greater detail on three main directions, including the background modeling based prediction techniques for high-efficiency surveillance video coding, error resilience methods for channel-noisy surveillance video transmission and the high-level syntax for surveillance video analysis. The surveillance groups can provide a good support for kinds of video analysis applications of computer vision and make the video transmission more applicable for noisy channels. Moreover, experimental results show that the background modeling based prediction techniques can well exploit the special characteristics of surveillance video and double the traditional compression performance.

Event detection in crowded surveillance videos is a challenging yet important problem. This paper focuses on pair-wise events that involve the interaction of two persons (e.g., people embrace, meet or split) in crowded videos. To detect such an event accurately, we should build an effective representation model that can characterize the sequential properties of two persons' interaction. Towards this end, we propose a novel pair-wise event detection approach using cubic features and sequence discriminant learning. A video sequence is first partitioned into several spatio-temporal cubes, and multiple features (e.g., statistics of trajectories, bag of spatio-temporal interest points) are extracted on these cubes and then fused to form a cubic feature descriptor under multiple kernel learning (MKL) framework. After that, the SVM with dynamic time alignment kernel is used to infer the existence of an event in the video sequence. Experimental results show that the proposed approach achieves the encouraging performance on TRECVid SED dataset.

The Deformable Part Model has shown high accuracy in tackling certain occlusion or deformations of objects such as cars and bikes. However, as for human category characterized by a larger number of articulated parts and more significant appearance variations, its performance gain is not so remarkable. To address this issue, we propose an MPLBoost-based mixture model which splits data into coherent groups and trains one root classifier for each, resulting in automated selection of discriminative root models and better representation of intra-class variations through visual feature clustering. Based on this boosting framework, multiple complementary features are combined to capture shape, texture and color information. Experimental results demonstrate that the proposed model can achieve an impressive performance improvement, especially in handling larger variations of human poses and viewpoints.

In the recent video coding standards, the selection of Lagrange multiplier is crucial to achieve trade-off between the choices of low-distortion and low-bitrate prediction modes. For surveillance video coding, the rate-distortion analysis shows that, a larger Lagrange multiplier should be used if the background in a coding unit took a larger proportion. Therefore, a modified Lagrange multiplier might be better for rate-distortion optimization. To address this problem, we perform an in-depth analysis on the relationship between the optimal Lagrange multiplier and the background proportion, and then propose a Lagrange multiplier selection model to obtain the optimal coding performance for surveillance videos. Following this, we further develop a Lagrange multiplier optimized video coding method. Experimental results show that our coding method can averagely achieve 18.07% bitrate saving on CIF sequences and 11.88% on SD sequences against the background-irrelevant Lagrange multiplier selection method.

As light is attenuated when disseminating in water, the clarity of images or videos captured under water is usually degraded to varying degrees. By exploring the difference in light attenuation between in atmosphere and in water, we derive a new underwater optical model to describe the formation of an underwater image in the true physical process, and then propose an effective enhancement algorithm with the derived optical model to improve the perception of underwater images or video frames. In our algorithm, a new underwater dark channel is derived to estimate the scattering rate, and an effective method is also presented to estimate the background light in the underwater optical model. Experimental results show that our algorithm can well handle underwater images, especially for deep-sea images and those captured from turbid waters.

To reduce the surveillance video coding cost, it is intuitive to encode surveillance videos by dealing with the foreground objects and the background separately. One widely used method following this strategy is Region-of-Interest (ROI) based coding. To achieve significant improvement for the coding efficiency of ROI based methods, this paper presents a surveillance video coding method with High Efficiency Video Coding (HEVC) quadtree partition based ROI extraction. With automatically generated foreground mask and modeled background frame, a ROI extraction following the block partition in HEVC's quadtree structure is firstly performed. Afterwards, surveillance videos can be compressed by coding two-layer videos. One is the ROI-layer video generated by merging ROIs and background data in each frame together. The other is the background-layer video produced by subtracting the ROIs from the original input video. Results show our method can achieve remarkable total bit-rate saving and significant bit-rate cost reduction on ROIs.

Smoke detection in video surveillance is very important for early fire detection. A general viewpoint assumes that smoke is a low frequency signal which may smoothen the background. However, some pure-color objects also have this characteristic, and smoke also produces high frequency signal because the rich edge information of its contour. In order to solve these problems, an improved smoke detection method with RGB Contrast-image and shape constrain is proposed. In this method, wavelet transformation is implemented based on the RGB Contrast-image to distinguish smoke from other low frequency signals, and the existence of smoke is determined by analyzing the combination of the shape and the energy change of the region. Experimental results show our method outperforms the conventional methods remarkably.

To save the storage and transmission cost, it is applicable now to develop fast and efficient methods to transcode the perennial surveillance videos to HEVC ones, since HEVC has doubled the compression ratio. Considering the long-time static background characteristic of surveillance videos, this paper presents a coding unit (CU) classification based AVC-to-HEVC transcoding method with background modeling. In our method, the background frame modeled from originally decoded frames is firstly transcoded into HEVC stream as long-term reference to enhance the prediction efficiency. Afterwards, a CU classification algorithm which employs decoded motion vectors and the modeled background frame as input is proposed to divide the decoded data into background, foreground and hybrid CUs. Following this, different transcoding strategies of CU partition termination, prediction unit candidate selection and motion estimation simplification are adopted for different CU categories to reduce the complexity. Experimental results show our method can achieve 45% bit saving and 50% complexity reduction against traditional AVC-to-HEVC transcoding.

This letter presents a novel approach for visual saliency estimation through single image optimization. Instead of directly mapping visual features to saliency values with a unified model, we treat regional saliency values as the optimization objective on each single image. By using a quadratic programming framework, our approach can adaptively optimize the regional saliency values on each specific image to simultaneously meet multiple saliency hypotheses on visual rarity, center-bias and mutual correlation. Experimental results show that our approach can outperform 14 state-of-the-art approaches on a public image benchmark.

Background subtraction is a fundamental preprocessing step in many surveillance video analysis tasks. In spite of significant efforts, however, background subtraction in crowded scenes remains challenging, especially, when a large number of foreground objects move slowly or just keep still. To address the problem, this paper proposes a selective eigenbackground method for background modeling and subtraction in crowded scenes. The contributions of our method are three-fold: First, instead of training eigenbackgrounds using the original video frames that may contain more or less foregrounds, a virtual frame construction algorithm is utilized to assemble clean background pixels from different original frames so as to construct some virtual frames as the training and update samples. This can significantly improve the purity of the trained eigenbackgrounds. Second, for a crowded scene with diversified environmental conditions (e.g., illuminations), it is difficult to use only one eigenbackground model to deal with all these variations, even using some online update strategies. Thus given several models trained offline, we utilize peak signal-to-noise ratio to adaptively choose the optimal one to initialize the online eigenbackground model. Third, to tackle the problem that not all pixels can obtain the optimal results when the reconstruction is performed at once for the whole frame, our method selects the best eigenbackground for each pixel to obtain an improved quality of the reconstructed background image. Extensive experiments on the TRECVID-SED dataset and the Road video dataset show that our method outperforms several state-of-the-art methods remarkably.

It is in urgent need to develop fast and efficient transcoding methods so as to remarkably save the storage of surveillance videos and synchronously transmit conference videos over different bandwidths. Towards this end, the special characteristics of these videos, e.g., the relatively static background, should be utilized for transcoding. Therefore, we propose a fast and efficient transcoding method (FET) based on background modeling and block classification in this paper. To improve the transcoding efficiency, FET adds the background picture, which is modeled from the originally decoded frames in low complexity, into stream in the form of an intra-coded G-picture. And then, FET utilizes the reconstructed G-picture as the long-term reference frame to transcode the following frames. This is mainly because our theoretical analyses show that G-picture can significantly improve the transcoding performance. To reduce the complexity, FET utilizes an adaptive threshold updating model for block classification and then adopts different transcoding strategies for different categories. This is due to the following statistics: after dividing blocks into categories of foreground, background and hybrid ones, different block categories have different distributions of prediction modes, motion vectors and reference frames. Extensive experiments on transcoding high-bit-rate H.264/AVC streams to low-bit-rate ones are carried out to evaluate our FET. Over the traditional full-decoding-and-full-encoding methods, FET can save more than 35% of the transcoding bit-rate with a speed-up ratio of larger than 10 on the surveillance videos. On the conference videos which should be transcoded more timely, FET achieves more than 20 times speed-up ratio with 0.2 dB gain.

Content-based copy detection (CBCD) is drawing increasing attention as an alternative technology to watermarking for video identification and copyright protection. In this article, we present a comprehensive method to detect copies that are subjected to complicated transformations. A multimodal feature representation scheme is designed to exploit the complementarity of audio features, global and local visual features so that optimal overall robustness to a wide range of complicated modifications can be achieved. Meanwhile, a temporal pyramid matching algorithm is proposed to assemble frame-level similarity search results into sequence-level matching results through similarity evaluation over multiple temporal granularities. Additionally, inverted indexing and locality sensitive hashing (LSH) are also adopted to speed up similarity search. Experimental results over benchmarking datasets of TRECVID 2010 and 2009 demonstrate that the proposed method outperforms other methods for most transformations in terms of copy detection accuracy. The evaluation results also suggest that our method can achieve competitive copy localization preciseness.

With the proliferation of moving surveillance cameras, how to effectively compress videos captured from them is becoming more and more important. One significant characteristic is that, these cameras always go and return cyclically within a limited area. Thus we propose to dynamically build up a background frame for each input frame from a generated panorama background and employ it for a background frame based motion compensation to improve the coding efficiency. For the background reconstruction procedure, we firstly extract limited number of feature point pairs between the robustly searched area in the decoded panorama and the current frame. Afterwards, the global motion transformation matrix is obtained to rectify the searched area into a projective plane of the current frame, and then the reconstructed background is produced. Experiments on six in-door and out-door surveillance videos show that, the background reconstruction based coding method achieves significant performance gain.

In this paper, we present our detection, tracking and event recognition methods and the results for PETS 2012. First, ROIs (Regions of Interest) based on geometric constraints are utilized in single view detection to eliminate the negative influence of clutter environment. Then, an optimized observation model is applied to address the ID switching or tracking drifting problem in single view tracking. Third, we introduce the multi-view Bayesian network (MBN) to reduce the "phantom" phenomena which frequently happen in general multi-view detection tasks. At last, a motion-based event recognition method is proposed to handle the event recognition task. Experimental results on the PETS 2012 dataset indicate that our methods are very promising.

Low-complexity and high-performance surveillance video Transcoding methods play an important role for a wide range of surveillance video transmission and storage applications. Towards this end, the special characteristics of surveillance video should be utilized for Transcoding. In this paper, we propose a fast and performance-maintained Transcoding method. This method firstly divides macro blocks (MBs) into foreground MBs, foreground border MBs and background MBs. Statistics show that the three categories have different distributions of prediction modes, motion vectors and reference frames. Following this, we adopt different Transco ding strategies in terms of removing the redundant prediction modes, narrowing motion search range and reducing reference frames. In particular, we propose an algorithm to exploit the decoded motion vector to adaptively calculate motion search range. Experimental results show that, compared with the recent background modeling based full-decoding-full-encoding, our Transcoding method saves more than 93% time with ignorable quality loss.

In the video copy detection task, it is widely recognized that none of any single feature can work well for all transformations. Thus more and more approaches adopt a set of complementary features to cope with complex audio-visual transformations. However, most of them utilize individual features separately and the final result is obtained by fusing results of several basic detectors. Often, this will lead to low detection efficiency. Moreover, there are some thresholds or parameters to be elaborately tuned. To address these problems, we propose a soft cascade approach to integrate multiple features for efficient copy detection. In our approach, basic detectors are organized in a cascaded framework, which processes a query video in sequence until one detector asserts it as a copy. To fully exert the complementarity of these detectors, a learning algorithm is proposed to estimate the optimal decision thresholds in the cascade architecture. Excellent performance on the benchmark dataset of TRECVid 2011 CBCD task demonstrates the effectiveness and efficiency of our approach.

Utilizing the special properties to improve the surveillance video coding efficiency still has much room, although there have been three typical paradigms of methods: object-oriented, background-prediction-based and background-difference-based methods. However, due to the inaccurate foreground segmentation, the low-quality or unclear background frame, and the potential "foreground pollution" phenomenon, there is still much room for improvement. To address this problem, this paper proposes a macro-block-level selective background difference coding method (MSBDC). MSBDC selects the following two ways to encode each macro-block (MB): coding the original MB, and directly coding the difference data between the MB and its corresponding background. MSBDC also features at employs the classification of MBs to facilitate the selection, through which, prediction and motion compensation turns more accurate, both on foreground and background. Results show that, MSBDC significantly decreases the total bitrate and obtains a remarkable performance gain on foreground compared with several state-of-the-art methods.

The goal of image authentication is to accept content-preserving operations and reject content-altering manipulations. So,it is increasingly approached by extracting content-based invariant features from original images and verifying their preservation in received images at later times. Since sparsity usually implies invariance, sparse feature representation has drawn significant attention from the research community. But only if discrimination is also found with a sparse feature, can it be successfully applied in image authentication. This paper proposes a sparse feature for image authentication by exploring the biologically-motivated standard model. Experimental results demonstrate both robustness and discrimination of the feature, and its effectiveness in tamper detection and location as well.

Background modeling is an important pre-processing step for object detection in surveillance video analysis systems. Recently, it has been proved to be useful for high-efficiency surveillance video coding. In existing works, the modeling background frame often needs to be high-quality encoded so as to achieve a large bit-rate saving. However, the high-quality background frame requires lots of bits in the code stream, so it is infeasible to update the background frame too frequently. Therefore, a better bit-allocation method is desirable to facilitate in-time background updating and bit-saving background coding. In this paper, we firstly build up a background updating model from a detailed analysis of results on surveillance video. Following this, we propose a bit-saving and quality-maintaining background frame coding method. In our method, the background frame can be updated more timely, consequently leading to the better coding efficiency. Experimental results show that our method can achieve more than 15% bit-rate decrease compared with three state-of-art methods.

Recently, background modeling (shortly BgModeling) plays a more and more important role in high-efficiency surveillance video coding. Meanwhile, many practical video coding applications also present some specific requirements for BgModeling, such as the low memory cost and low computational complexity. However, existing BgModeling methods are mostly designed for video content analysis such as object detection. Thus they may be not directly applicable for video coding. In this paper, we firstly present an analysis for the features of BgModeling in surveillance video coding and make a comparison of the performances of existing BgModeling methods. Then we propose a segment-and-weight based running average (SWRA) method for surveillance video coding. SWRA firstly divides pixels at each position in the training frames into several temporal segments, and then calculate their corresponding mean values and weights. After that, a running and weighted average procedure is used to reduce the influence of foreground pixels and finally obtain the modeling results. Experimental results show that, the SWRA-based encoder achieves the best performance over several state-of-the-art methods, with much less cost of memory and modeling time.

The advance of the Internet in the past decade has radically changed the way people communicate and collaborate with each other. Physical distance is no more a barrier in online social networks, but cultural differences (at the individual, community, as well as societal levels) still govern human-human interactions and must be considered and leveraged in the online world. The rapid deployment of high-speed Internet allows humans to interact using a rich set of multimedia data such as texts, pictures, and videos. This position paper proposes to define a new research area called ‘connected multimedia’, which is the study of a collection of research issues of the super-area social media that receive little attention in the literature. By connected multimedia, we mean the study of the social and technical interactions among users, multimedia data, and devices across cultures and explicitly exploiting the cultural differences. We justify why it is necessary to bring attention to this new research area and what benefits of this new research area may bring to the broader scientific research community and the humanity.

In this paper, a group-sensitive multiple kernel learning (GS-MKL) method is proposed for object recognition to accommodate the intraclass diversity and the interclass correlation. By introducing the "group" between the object category and individual images as an intermediate representation, GS-MKL attempts to learn group-sensitive multikernel combinations together with the associated classifier. For each object category, the image corpus from the same category is partitioned into groups. Images with similar appearance are partitioned into the same group, which corresponds to the subcategory of the object category. Accordingly, intraclass diversity can be represented by the set of groups from the same category but with diverse appearances; interclass correlation can be represented by the correlation between groups from different categories. GS-MKL provides a tractable solution to adapt multikernel combination to local data distribution and to seek a tradeoff between capturing the diversity and keeping the invariance for each object category. Different from the simple hybrid grouping strategy that solves sample grouping and GS-MKL training independently, two sample grouping strategies are proposed to integrate sample grouping and GS-MKL training. The first one is a looping hybrid grouping method, where a global kernel clustering method and GS-MKL interact with each other by sharing group-sensitive multikernel combination. The second one is a dynamic divisive grouping method, where a hierarchical kernel-based grouping process interacts with GS-MKL. Experimental results show that performance of GS-MKL does not significantly vary with different grouping strategies, but the looping hybrid grouping method produces slightly better results. On four challenging data sets, our proposed method has achieved encouraging performance comparable to the state-of-the-art and outperformed several existing MKL methods.

Image authentication is usually approached by checking the preservation of some invariant features, which are expected to be both robust and discriminative so that content-preserving operations are accepted while content-altering manipulations are rejected. However, most of existing features have not obtained convincing performance due to insufficiency of experiments and over biasing of robustness. Motivated by the sparse coding strategy discovered in primary visual cortex, we explore the possibility of using sparse coding coefficients for image authentication. Through extensive experiments, we discover that the proposed feature bears great discrimination as well as robustness, which indicates the effectiveness of sparse coding as a new invariant feature for image authentication.

In this paper, a selective eigenbackgrounds method is proposed for background subtraction in crowded scenes. In order to train and update the eigenbackground model with frames containing few objects (i.e. clean frames), virtual frames are constructed based on a frame selection map. Then, the eigenbackground that best depicts background is selected for each pixel based on an eigenbackground selection map. Experimental results show the performance of the proposed method is better than those of some state-of-the-art methods in crowded scenes.

Content-based video copy detection over large corpus with complex transformations is important but challenging. It is not surprising that most existing methods fall short of either sufficient robustness to detect severely deformed copies or high accuracy to localize copy segments. In this paper, we propose a video copy detection approach which exploits complementary audio-visual features and sequential pyramid matching (SPM). Several independent detectors first match visual key frames or audio clips using individual features, and then aggregate the frame level results into video level results with SPM, which calculates video similarities by sequence matching at multiple granularities. Finally, detection results from basic detectors are fused and further filtered to generate the final result. Excellent performance evaluated on TRECVid 2010 copy detection task demonstrates the effectiveness of our approach.

Visual saliency plays an important role in various video applications such as video retargeting and intelligent video advertising. However, existing visual saliency estimation approaches often construct a unified model for all scenes, thus leading to poor performance for the scenes with diversified contents. To solve this problem, we propose a multi-task rank learning approach which can be used to infer multiple saliency models that apply to different scene clusters. In our approach, the problem of visual saliency estimation is formulated in a pair-wise rank learning framework, in which the visual features can be effectively integrated to distinguish salient targets from distractors. A multi-task learning algorithm is then presented to infer multiple visual saliency models simultaneously. By an appropriate sharing of information across models, the generalization ability of each model can be greatly improved. Extensive experiments on a public eye-fixation dataset show that our multi-task rank learning approach outperforms 12 state-of-the-art methods remarkably in visual saliency estimation.

In this paper, we present our eSur (Event detection system on SURveillance video) system, which is derived from TRECVID'09 surveillance tasks. Currently, eSur attempts to detect two categories of events: 1) single-actor events (i.e., PersonRuns and ElevatorNoEntry) irrespective of any interaction between individuals, and 2) pair-activity events (i.e., PeopleMeet, PeopleSplitUp, and Embrace) involves more than one individual. eSur consists of three major stages, i.e., preprocessing, event classification, and post-processing. The preprocessing involves view classification, background subtraction, head-shoulder detection, human body detection and object tracking. Event classification fuses One-vs.-All SVM and rule-based classifiers to identify single-actor and pair-activity events in an ensemble way. To reduce false alarms, we introduce prior knowledge into the post-processing, and in particular, we apply a so-called event merging process over TRECVID dataset. Extensive experiments have been performed over TRECVid'08 and '09 ED data corpus involving in total 144 hours surveillance video of London Gatwick airport. According to the TRECVid-ED formal evaluation, our prototype has yielded fairly promising results over TRECVid'09 dataset, with top Act.DCR of 1.023, 1.025, 1.02, and 0.334 for PeopleMeet, PeopleSplitUp, Embrace, and ElevatorNoEntry, respectively.

Multi-cue integration has proved successful at increasing the robustness of tracking algorithms and overcoming the failure cases of individual cue. But considering dynamic appearance of objects or clutter background, the integration based on constant weights may weaken the performance of this scheme. In this paper, we propose a dynamic weights update mechanism for multiple cues tracking with detection responses as supervision. We integrate multiple cues based on the observation hypotheses compared with detection association results and adjust the weights according to the approximation degree. The integration is adapted on-the-fly during tracking, in order to keep the tracker adaptive. The proposed method allows flexible combination of different cues and we select cues based on color and local feature for tracking. Experiments are carried out on 602 trajectories extracted from TRECVID 2008 event detection dataset which is recorded in an airport scenario. Comparison results prove the effectiveness of our method.

Visual saliency can be a useful tool for image content analysis such as automatic image cropping and image compression. In existing methods on visual saliency detection, most of them are related to the model of receptive field. In this paper, we propose a bottom-up model which introduces 2D Log-Gabor wavelets for saliency detection. Compared with the traditional model of receptive field, the 2D Log-Gabor wavelets can better simulate the biological characteristics of the simple cortical cell in the receptive filed. Moreover, we also incorporate the influence of center bias into our model, which is a common phenomenon that directs visual attention to the center of images in natural scenes. Experimental results show that our approach outperforms three state-of-the-art approaches remarkably.

Automatic interesting object extraction is widely used in many image applications. Among various extraction approaches, saliency-based ones usually have a better performance since they well accord with human visual perception. However, nearly all existing saliency-based approaches suffer the integrity problem, namely, the extracted result is either a small part of the object (referred to as sketch-like) or a large region that contains some redundant part of the background (referred to as envelope-like). In this paper, we propose a novel object extraction approach by integrating two kinds of "complementary" saliency maps (i.e., sketch-like and envelope-like maps). In our approach, the extraction process is decomposed into two sub-processes, one used to extract a high-precision result based on the sketch-like map, and the other used to extract a high-recall result based on the envelope-like map. Then a classification step is used to extract an exact object based on the two results. By transferring the complex extraction task to an easier classification problem, our approach can effectively break down the integrity problem. Experimental results show that the proposed approach outperforms six state-of-art saliency-based methods remarkably in automatic object extraction, and is even comparable to some interactive approaches.

Recently, we designed a new experimental system MSearch, which is a cross-media meta-search system built on the database of the WikipediaMM task of ImageCLEF 2008. For a meta-search engine, the kernel problem is how to merge the results from multiple member search engines and provide a more effective rank list. This paper deals with a novel fusion model employing supervised learning. Our fusion model employs ranking SVM in training the fusion weight for each member search engine. We assume the fusion weight of each member search engine as a feature of a result document returned by the meta-search engine. For a returned result document, we first build a feature vector to represent the document, and set the value of each feature as the document’s score returned by the corresponding member search engine. Then we construct a training set from the documents returned from the meta-search engine to learn the fusion parameter. Finally, we use the linear fusion model based on the overlap set to merge the results set. Experimental results show that our approach significantly improves the performance of the cross-media meta-search (MSearch) and outperforms many of the existing fusion methods.

The increasing amount of videos on the Internet and digital libraries highlights the necessity and importance of interactive video services such as automatically associating additional materials (e.g., advertising logos and relevant selling information) with the video content so as to enrich the viewing experience. Toward this end, this paper presents a novel approach for user-targeted video content association (VCA). In this approach, the salient objects are extracted automatically from the video stream using complementary saliency maps. According to these salient objects, the VCA system can push the related logo images to the users. Since the salient objects often correspond to important video content, the associated images can be considered as content-related. Our VCA system also allows users to associate images to the preferred video content through simple interactions by the mouse and an infrared pen. Moreover, by learning the preference of each user through collecting feedbacks on the pulled or pushed images, the VCA system can provide user-targeted services. Experimental results show that our approach can effectively and efficiently extract the salient objects. Moreover, subjective evaluations show that our system can provide content-related and user-targeted VCA services in a less intrusive way.

This paper presents a cost-sensitive rank learning approach for visual saliency estimation. This approach avoids the explicit selection of positive and negative samples, which is often used by existing learning-based visual saliency estimation approaches. Instead, both the positive and unlabeled data are directly integrated into a rank learning framework in a cost-sensitive manner. Compared with existing approaches, the rank learning framework can take the influences of both the local visual attributes and the pair-wise contexts into account simultaneously. Experimental results show that our algorithm outperforms several state-of-the-art approaches remarkably in visual saliency estimation.

Automatic video annotation is a challenging yet important problem for content-based video indexing and retrieval. In most existing works, annotation is formulated as a multi-labeling problem over individual shots. However, video is by nature informative in spatial and temporal context of semantic concepts. In this paper, we formulate video annotation as a sequence multi-labeling (SML) problem over a shot sequence. Different from many video annotation paradigms working on individual shots, SML aims to predict a multi-label sequence for consecutive shots in a global optimization manner by incorporating spatial and temporal context into a unified learning framework. A novel discriminative method, called sequence multi-label support vector machine (SVMSML), is accordingly proposed to infer the multi-label sequence for a given shot sequence. In SVMSML, a joint kernel is employed to model the feature-level and concept-level context relationships (i.e., the dependencies of concepts on the low-level features, spatial and temporal correlations of concepts). A multiple-kernel learning (MKL) algorithm is developed to optimize the kernel weights of the joint kernel as well as the SML score function. To efficiently search the desirable multi-label sequence over the large output space in both training and test phases, we adopt an approximate method to maximize the energy of a binary Markov random field (BMRF). Extensive experiments on TRECVID'05 and TRECVID'07 datasets have shown that our proposed SVMSMLgains superior performance over the state-of-the-art.

In this paper, we propose a group-sensitive multiple kernel learning (GS-MKL) method to accommodate the intra-class diversity and the inter-class correlation for object categorization. By introducing an intermediate representation “group” between images and object categories, GS-MKL attempts to find appropriate kernel combination for each group to get a finer depiction of object categories. For each category, images within a group share a set of kernel weights while images from different groups may employ distinct sets of kernel weights. In GS-MKL, such group-sensitive kernel combinations together with the multi-kernels based classifier are optimized in a joint manner to seek a trade-off between capturing the diversity and keeping the invariance for each category. Extensive experiments show that our proposed GS-MKL method has achieved encouraging performance over three challenging datasets.

Recently, visual saliency has drawn great research interest in the field of computer vision and multimedia. Various approaches aiming at calculating visual saliency have been proposed. To evaluate these approaches, several datasets have been presented for visual saliency in images. However, there are few datasets to capture spatiotemporal visual saliency in video. Intuitively, visual saliency in video is strongly affected by temporal context and might vary significantly even in visually similar frames. In this paper, we present an extensive dataset with 7.5-hour videos to capture spatiotemporal visual saliency. The salient regions in frames sequentially sampled from these videos are manually labeled by 23 subjects and then averaged to generate the ground-truth saliency maps. We also present three metrics to evaluate competing approaches. Several typical algorithms were evaluated on the dataset. The experimental results show that this dataset is very suitable for evaluating visual saliency. We also discover some interesting findings that would be addressed in future research. Currently, the dataset is freely available online together with the source code for evaluation.

Recently, multiple kernel learning (MKL) methods have shown promising performance in image classification. As a sort of supervised learning, training MKL-based classifiers relies on selecting and annotating extensive dataset. In general, we have to manually label large amount of samples to achieve desirable MKL-based classifiers. Moreover, MKL also suffers a great computational cost on kernel computation and parameter optimization. In this paper, we propose a local adaptive active learning (LA-AL) method to reduce the labeling and computational cost by selecting the most informative training samples. LA-AL adopts a top-down (or global-local) strategy for locating and searching informative samples. Uncertain samples are first clustered into groups, and then informative samples are consequently selected via inter-group and intra-group competitions. Experiments over COREL-5K show that the proposed LA-AL method can significantly reduce the demand of sample labeling and have achieved the state-of-the-art performance.

Text segmentation, or named text binarization, is usually an essential step for text information extraction from images and videos. However, most existing text segmentation methods have difficulties in extracting multi-polarity texts, where multi-polarity texts mean those texts with multiple colors or intensities in the same line. In this paper, we propose a novel algorithm for multi-polarity text segmentation based on graph theory. By representing a text image with an undirected weighted graph and partitioning it iteratively, multi-polarity text image can be effectively split into several single-polarity text images. As a result, these text images are then segmented by single-polarity text segmentation algorithms. Experiments on thousands of multi-polarity text images show that our algorithm can effectively segment multi-polarity texts.

Generally speaking, digital libraries have multiple granularities of semantic units: book, chapter, page, paragraph and word. However, there are two limitations of current eBook retrieval systems: (1) the granularity of retrievable units is either too big or too small, scales such as chapters, paragraphs are ignored; (2) the retrieval results should be grouped by facets to facilitate user's browsing and exploration. To overcome these limitations, we propose a multi-granularity multi-facet eBook retrieval approach.

Keyphrases play a key role in text indexing, summarization and categorization. However, most of the existing keyphrase extraction approaches require human-labeled training sets. In this paper, we propose an automatic keyphrase extraction algorithm, which can be used in both supervised and unsupervised tasks. This algorithm treats each document as a semantic network. Structural dynamics of the network are used to extract keyphrases (key nodes) unsupervised. Experiments demonstrate the proposed algorithm averagely improves 50% in effectiveness and 30% in efficiency in unsupervised tasks and performs comparatively with supervised extractors. Moreover, by applying this algorithm to supervised tasks, we develop a classifier with an overall accuracy up to 80%.

In the area of image retrieval, post-retrieval processing is often used to refine the retrieval results to better satisfy users' requirements. Previous methods mainly focus on presenting users with relevant results. However, in most cases, users cannot clearly present their requirements by several query words. Therefore, relevant results with rich topic coverage are more likely to meet users' ambiguous needs. In this paper, a re-ranking method based on topic richness analysis is proposed to enrich topic coverage in retrieval results. Furthermore, a quantitative criterion called diversity scores (DS) is proposed to evaluate the improvement. Given a set of images, topics that are rarely included in the set are scarce topics, as oppose to rich topics that are widely distributed among the set. Scarce topics contribute more than rich topics do to the DS of images. Five researchers are invited to evaluate the re-ranked results both in topic coverage and relevance. Experimental results on over 20,000 images demonstrate that our proposed approach is effective in improving the topic coverage of retrieval results without loss of relevance.

Links among objects contain rich semantics that can be very helpful in classifying the objects. However, many irrelevant links can be found in real-world link data such as Web pages. Often, these noisy and irrelevant links do not provide useful and predictive information for categorization. It is thus important to automatically identify which links are most relevant for categorization. In this paper, we present a contextual dependency network (CDN) model for classifying linked objects in the presence of noisy and irrelevant links. The CDN model makes use of a dependency function that characterizes the contextual dependencies among linked objects. In this way, CDNs can differentiate the impacts of the related objects on the classification and consequently reduce the effect of irrelevant links on the classification. We show how to learn the CDN model effectively and how to use the Gibbs inference framework over the learned model for collective classification of multiple linked objects. The experiments show that the CDN model demonstrates relatively high robustness on data sets containing irrelevant links.

With the exponential growth of both the amount and diversity of the information that the Web encompasses, automatic classification of topic-specific Web sites is highly desirable. We propose a novel approach for Web site classification based on the content, structure and context information of Web sites. In our approach, the site structure is represented as a two-layered tree in which each page is modeled as a DOM (document object model) tree and a site tree is used to hierarchically link all pages within the site. Two context models are presented to capture the topic dependences in the site. Then the hidden Markov tree (HMT) model is utilized as the statistical model of the site tree and the DOM tree, and an HMT-based classifier is presented for their classification. Moreover, for reducing the download size of Web sites but still keeping high classification accuracy, an entropy-based approach is introduced to dynamically prune the site trees. On these bases, we employ the two-phase classification system for classifying Web sites through a fine-to-coarse recursion. The experiments show our approach is able to offer high accuracy and efficient process performance.