Knowledge Distillation (KD) is a key technique for compressing Large-scale Language Models (LLMs), but prevailing logit-based methods employ static strategies misaligned with the student’s dynamic learning process. By treating all tokens indiscriminately with a fixed temperature, these methods result in suboptimal knowledge transfer. To address this, we propose LLM-oriented token-Adaptive Knowledge Distillation (AdaKD), a framework that adapts the distillation process to each token’s real-time learning state. AdaKD consists of two synergistic modules driven by a unified token difficulty metric. First, the Loss-driven Adaptive Token Focusing (LATF) module dynamically concentrates distillation on valuable tokens by monitoring the student’s learning stability. Second, Inverse Difficulty Temperature Scaling (IDTS) introduces a counterintuitive token-level temperature: low for difficult tokens to target error correction, and high for easy tokens to learn the teacher’s smooth output distribution for better generalization. As a plug-and-play framework, AdaKD consistently improves performance across diverse distillation methods, model architectures, and benchmarks.

Watermarking diffusion-generated images is crucial for copyright protection and user tracking. However, current diffusion watermarking methods face significant limitations: zero-bit watermarking systems lack the capacity for large-scale user tracking, while multi-bit methods are highly sensitive to certain image transformations or generative attacks, resulting in a lack of comprehensive robustness. In this paper, we propose OptMark, an optimization-based approach that embeds a robust multi-bit watermark into the intermediate latents of the diffusion denoising process. OptMark strategically inserts a structural watermark early to resist generative attacks and a detail watermark late to withstand image transformations, with tailored regularization terms to preserve image quality and ensure imperceptibility. To address the challenge of memory consumption growing linearly with the number of denoising steps during optimization, OptMark incorporates adjoint gradient methods, reducing memory usage from O(N) to O(1). Experimental results demonstrate that OptMark achieves invisible multi-bit watermarking while ensuring robust resilience against valuemetric transformations, geometric transformations, editing, and regeneration attacks.

Large Reasoning Models (LRMs) with long chain-of-thought reasoning have recently achieved remarkable success.Yet, equipping domain-specialized models with such reasoning capabilities, referred to as "Reasoning + X", remains a significant challenge.While model merging offers a promising training-free solution, existing methods often suffer from a destructive performance collapse: existing methods tend to both weaken reasoning depth and compromise domain-specific utility.Interestingly, we identify a counter-intuitive phenomenon underlying this failure: reasoning ability predominantly resides in parameter regions with low gradient sensitivity, contrary to the common assumption that domain capabilities correspond to high-magnitude parameters.Motivated by this insight, we propose ReasonAny, a novel merging framework that resolves the reasoning-domain performance collapse through Contrastive Gradient Identification.Experiments across safety, biomedicine, and finance domains show that ReasonAny effectively synthesizes "Reasoning + X" capabilities, significantly outperforming state-of-theart baselines while retaining robust reasoning performance.

Open-vocabulary 3D scene understanding presents a significant challenge in the field. Recent works have sought to transfer knowledge embedded in vision-language models from 2D to 3D domains. However, these approaches often require prior knowledge from specific 3D scene datasets, limiting their applicability in open-world scenarios. The Segment Anything Model (SAM) has demonstrated remarkable zero-shot segmentation capabilities, prompting us to investigate its potential for comprehending 3D scenes without training. In this paper, we introduce OV-SAM3D, a training-free method that contains a universal framework for understanding open-vocabulary 3D scenes. This framework is designed to perform understanding tasks for any 3D scene without requiring prior knowledge of the scene. Specifically, our method is composed of two key sub-modules: First, we initiate the process by generating superpoints as the initial 3D prompts and refine these prompts using segment masks derived from SAM. Moreover, we then integrate a specially designed overlapping score table with open tags from the Recognize Anything Model (RAM) to produce final 3D instances with open-world labels. Empirical evaluations on the ScanNet200 and nuScenes datasets demonstrate that our approach surpasses existing open-vocabulary methods in unknown open-world environments. Code is available at https://github.com/HanchenTai/OV-SAM3D.git.

Accurate segmentation of cervical structures in transvaginal ultrasound (TVS) is critical for assessing the risk of spontaneous preterm birth (PTB), yet the scarcity of labeled data limits the performance of supervised learning approaches. This paper introduces the Fetal Ultrasound Grand Challenge (FUGC), the first benchmark for semi-supervised learning in cervical segmentation, hosted at ISBI 2025. FUGC provides a dataset of 890 TVS images, including 500 training images, 90 validation images, and 300 test images. Methods were evaluated using the Dice Similarity Coefficient (DSC), Hausdorff Distance (HD), and runtime (RT), with a weighted combination of 0.4/0.4/0.2. The challenge attracted 10 teams with 82 participants submitting innovative solutions. The best-performing methods for each individual metric achieved 90.26% mDSC, 38.88 mHD, and 32.85 ms RT, respectively. FUGC establishes a standardized benchmark for cervical segmentation, demonstrates the efficacy of semi-supervised methods with limited labeled data, and provides a foundation for AI-assisted clinical PTB risk assessment.

Co-speech gesture generation plays a vital role in producing synchronized and natural human gestures, thereby enhancing the realism of avatars in virtual environments. Although diffusion models have shown strong generative capabilities, their combination with transformer-based architectures often incurs high computational costs due to the quadratic complexity of self-attention. Moreover, as a temporal sequence modeling task, existing methods frequently struggle to effectively capture multi-scale temporal dynamics inherent in speech and gesture signals. To address these challenges, we propose MambaGesture2, a novel framework that integrates a Mamba-based denoising network, Hierarchical U-Net Gesture Mamba (HUG-Mamba), with a multimodal feature fusion module, SEAD. HUG-Mamba combines the efficient state-space modeling of Mamba blocks with the hierarchical sampling of the U-Net architecture, significantly improving temporal coherence and computational efficiency. We further introduce the Temporal-Stratified Fusion (TSF) module to capture diverse temporal scales via multi-scale learning, and the Spatial-Temporal Cascaded Aggregation (STCA) module to enhance spatial-temporal feature aggregation. Extensive experiments on the multi-modal BEAT2 and SHOW datasets demonstrate that our approach achieves state-of-the-art performance across multiple quantitative metrics, while substantially reducing model complexity and inference time. The results validate the effectiveness of our architectural innovations in generating diverse, realistic, and temporally consistent co-speech gestures. Project page:https://fcchit.github.io/mambagesture2.

Source-free domain adaptation (SFDA) is a challenging, yet valuable task within unsupervised domain adaptation (UDA), which adapts pretrained models to diverse unlabeled target domains while safeguarding the data security of the source domain. However, existing SFDA methods primarily focus on computer vision applications, often overlooking the unique characteristics of time series, such as temporal dependencies and sequential nature. Moreover, the fine-tuning paradigm of current SFDA methods is typically limited to posterior adaptation, focusing solely on constraining the statistical properties of model outputs. We argue that this black-box paradigm lacks semantic interpretability and risks aligning with spurious contextual noise, leading to negative transfer. This necessitates a paradigm evolution from blind statistical adaptation to interpretable adaptation. To this end, we introduce model salience as a quantifiable proxy of semantic interpretability, representing the importance weights a trained model assigns to specific temporal fragments. Accordingly, we propose a novel fine-tuning paradigm for time-series SFDA, termed PrEPoA, which integrates Prior Evaluation of model salience with Posterior Adaptation. In the prior evaluation stage, a key pattern reconstruction (KPR) module based on a sensitive masking mechanism is designed to quantify the model salience, while a novel interpattern triplet loss is introduced to calibrate it. In the posterior adaptation stage, robust prototype clustering (RPC) generates trustworthy reference labels as pseudo-ground truth for adaptation. Comprehensive experiments on the wireless sensor data mining (WISDM), human activity recognition (HAR), heterogeneity HAR (HHAR), machine fault diagnosis (MFD), and sleep stage classification (SSC) datasets demonstrate the superiority of our PrEPoA framework compared to nine UDA and seven SFDA methods. Furthermore, we experimentally validate that PrEPoA serves as a plug-and-play module that effectively incorporated into other SFDA methods.

While autonomous driving technology has made remarkable strides, data-driven approaches still struggle with complex scenarios due to their limited reasoning capabilities. Meanwhile, knowledge-driven autonomous driving systems have evolved considerably with the popularization of visual language models. In this article, we propose LeapVAD, a novel method based on cognitive perception and dual-process thinking. Our approach implements a human-attentional mechanism to identify and focus on critical traffic elements that influence driving decisions. By characterizing these objects through comprehensive attributes-including appearance, motion patterns, and associated risks-LeapVAD achieves more effective environmental representation and streamlines the decision-making process. Furthermore, LeapVAD incorporates an innovative dual-process decision-making module mimicking the human-driving learning process. The system consists of an analytic process (System-II) that accumulates driving experience through logical reasoning and a heuristic process (System-I) that refines this knowledge via fine-tuning and few-shot learning. LeapVAD also includes reflective mechanisms and a growing memory bank, enabling it to learn from past mistakes and continuously improve its performance in a closed-loop environment. To enhance efficiency, we develop a scene encoder network that generates compact scene representations for rapid retrieval of relevant driving experiences. Extensive evaluations conducted on two leading autonomous driving simulators, CARLA and DriveArena, demonstrate that LeapVAD achieves superior performance compared with camera-only approaches despite limited training data. Comprehensive ablation studies further emphasize its effectiveness in continuous learning and domain adaptation. Project page: https://pjlab-adg.github.io/LeapVAD/.

We propose Transmission-Reflection Gaussians (TR-Gaussians), a novel 3D-Gaussian-based representation for high-fidelity rendering of planar transmission and reflection, which are ubiquitous in indoor scenes. Our method combines 3D Gaussians with learnable reflection planes that explicitly model the glass planes with view-dependent reflectance strengths. Real scenes and transmission components are modeled by 3D Gaussians and the reflection components are modeled by the mirrored Gaussians with respect to the reflection plane. The transmission and reflection components are blended according to a Fresnel-based, view-dependent weighting scheme, allowing for faithful synthesis of complex appearance effects under varying viewpoints. To effectively optimize TR-Gaussians, we develop a multi-stage optimization framework incorporating color and geometry constraints and an opacity perturbation mechanism. Experiments on different datasets demonstrate that TR-Gaussians achieve real-time, high-fidelity novel view synthesis in scenes with planar transmission and reflection, and outperform state-of-the-art approaches both quantitatively and qualitatively.

This work aims at a challenging task: human action reaction synthesis, i.e., generating human reactions conditioned on the action sequence of another person. Currently, autoregressive modeling approaches with vector quantization (VQ) have achieved remarkable performance in motion generation tasks. However, VQ has inherent disadvantages, including quantization information loss, low codebook utilization, etc. In addition, while dividing the body into separate units can be beneficial, the computational complexity needs to be considered. Also, the importance of mutual perception among units is often neglected. In this work, we propose MARRS, a novel framework designed to generate coordinated and fine-grained reaction motions us ing continuous representations. Initially, we present the Unit distinguished Motion Variational AutoEncoder (UD-VAE), which segments the entire body into distinct body and hand units, encoding each independently. Subsequently, we propose Action Conditioned Fusion (ACF), which involves randomly masking a subset of reactive tokens and extracting specific information about the body and hands from the active tokens. Furthermore, we introduce Mutual Unit Modulation (MUM) to facilitate interaction between body and hand units by using the information from one unit to adaptively modulate the other. Finally, for the diffusion model, we employ a compact MLP as a noise predictor for each distinct body unit and incorporate the diffusion loss to model the probability distribution of each token. Both quantitative and qualitative results demonstrate that our method achieves superior performance. The code will be released upon acceptance.

World models envision potential future states based on various ego actions. They embed extensive knowledge about the driving environment, facilitating safe and scalable autonomous driving. Most existing methods primarily focus on either data generation or the pretraining paradigms of world models. Unlike the aforementioned prior works, we propose Drive-OccWorld, which adapts a vision-centric 4D forecasting world model to end-to-end planning for autonomous driving. Specifically, we first introduce a semantic and motion-conditional normalization in the memory module, which accumulates semantic and dynamic information from historical BEV embeddings. These BEV features are then conveyed to the world decoder for future occupancy and flow forecasting, considering both geometry and spatiotemporal modeling. Additionally, we propose injecting flexible action conditions, such as velocity, steering angle, trajectory, and commands, into the world model to enable controllable generation and facilitate a broader range of downstream applications. Furthermore, we explore integrating the generative capabilities of the 4D world model with end-to-end planning, enabling continuous forecasting of future states and the selection of optimal trajectories using an occupancy-based cost function. Extensive experiments on the nuScenes dataset demonstrate that our method can generate plausible and controllable 4D occupancy, opening new avenues for driving world generation and end-to-end planning.

ENT (Ear, Nose, and Throat) endoscopic imaging plays a crucial role in the diagnosis and treatment planning of a wide range of head and neck disorders, yet remains underrepresented in the development of automated image analysis methods. While endoscopic image classification has been extensively explored in various anatomical domains, limited attention has been paid to the ear, nose, and throat (ENT) region. To address this gap, we conduct a systematic study on ENT endoscopy image classification, motivated by our 3rdplace solution in the ENTRep Challenge at ACM Multimedia 2025. We train a total of 190 models, spanning 19 representative architectures from both convolutional neural networks (CNNs) and vision transformers (ViTs), under a 10-fold cross-validation setting. In addition, we propose a selective ensemble strategy, which achieves improved classification performance with reduced computational cost. Additionally, we provide an empirical analysis of training strategies, including the effects of pretraining and data augmentation, offering practical insights into effective model design for ENT applications. This work represents a systematic evaluation of ENT endoscopic image classification methods and provides a strong baseline and methodological guidance for future research in this domain.

Multimodal Language Learning Models (MLLMs) have shown remarkable performance in image understanding, generation, and editing, with recent advancements achieving pixel-level grounding with reasoning. However, these models for common objects struggle with fine-grained face understanding. In this work, we introduce the FacePlayGround-240K dataset, the first pioneering large-scale, pixel-grounded face caption and question-answer (QA) dataset that includes 240K images, 47 mask categories, 5.4M mask annotations, and 7.3M grounded regions, meticulously curated for alignment pretraining and instruction-tuning. We present the GroundingFace framework, specifically designed to enhance fine-grained face understanding. This framework significantly augments the capabilities of existing grounding models in face part segmentation, face attribute comprehension, while preserving general scene understanding. Comprehensive experiments validate that our approach surpasses current state-of-the-art models in pixel-grounded face captioning/QA and various downstream tasks, including face captioning, referring segmentation, and zero-shot face attribute recognition.

Video action recognition involves interpreting both global context and specific details to accurately identify actions. While previous models are effective at capturing spatiotemporal features, they often lack a focused representation of key action details. To address this, we introduce FocusVideo, a framework designed for refining video action recognition through integrated global and local feature learning. Inspired by human visual cognition theory, our approach balances the focus on both broad contextual changes and action-specific details, minimizing the influence of irrelevant background noise. We first employ learnable action queries to selectively emphasize action-relevant regions without requiring region-specific labels. Next, these queries are learned by a local action streaming branch that enables progressive query propagation. Moreover, we introduce a parameter-free feature interaction mechanism for effective multi-scale interaction between global and local features with minimal additional overhead. Extensive experiments demonstrate that FocusVideo achieves state-of-the-art performance across multiple action recognition datasets, validating its effectiveness and robustness in handling action-relevant details.

Human-human motion generation is essential for understanding humans as social beings. Current methods fall into two main categories: single-person-based methods and separate modeling-based methods. To delve into this field, we abstract the overall generation process into a general framework MetaMotion, which consists of two phases: temporal modeling and interaction mixing. For temporal modeling, the single-person-based methods concatenate two people into a single one directly, while the separate modeling-based methods skip the modeling of interaction sequences. The inadequate modeling described above resulted in sub-optimal performance and redundant model parameters. In this paper, we introduce TIMotion (Temporal and Interactive Modeling), an efficient and effective framework for human-human motion generation. Specifically, we first propose Causal Interactive Injection to model two separate sequences as a causal sequence leveraging the temporal and causal properties. Then we present Role-Evolving Scanning to adjust to the change in the active and passive roles throughout the interaction. Finally, to generate smoother and more rational motion, we design Localized Pattern Amplification to capture short-term motion patterns. Extensive experiments on InterHuman and Inter-X demonstrate that our method achieves superior performance.

Real-world scenes likely involve repetitive objects indicating that the reconstruction of the target object can be supplemented by the views of other identical objects. However, traditional 3D reconstruction methods do not take this a priori knowledge into account and fail to make full use of the available information. In this paper, we propose an object-aware viewpoint augmentation scheme for indoor compositional reconstruction. Within this scheme, a viewpoint supplementation strategy based on signed distance function and neural radiance fields is proposed to fully leverage the information from repetitive objects such that the occlusion problem is suppressed. Moreover, this scheme introduces monocular uncertainty priors and regional smoothness constraints to enhance the reconstruction accuracy of slender and thin structures and the smoothness of occluded background, respectively. Experimental results considering both synthetic and real-world scenes demonstrate that our method effectively improves the reconstruction quality of repetitive objects and background.

Video summarization, by selecting the most informative and/or user-relevant parts of original videos to create concise summary videos, has high research value and consumer demand in today’s video proliferation era. Multi-modal video summarization that accomodates user input has become a research hotspot. However, current multi-modal video summarization methods suffer from two limitations. First, existing methods inadequately fuse information from different modalities and cannot effectively utilize modality-unique features. Second, most multi-modal methods focus on video and text modalities, neglecting the audio modality, despite the fact that audio information can be very useful in certain types of videos. In this paper we propose CFSum, a transformer-based multi-modal video summarization framework with coarse-fine fusion. CFSum exploits video, text, and audio modal features as input, and incorporates a two-stage transformer-based feature fusion framework to fully utilize modality-unique information. In the first stage, multi-modal features are fused simultaneously to perform initial coarse-grained feature fusion, then, in the second stage, video and audio features are explicitly attended with the text representation yielding more fine-grained information interaction. The CFSum architecture gives equal importance to each modality, ensuring that each modal feature interacts deeply with the other modalities. Our extensive comparative experiments against prior methods and ablation studies on various datasets confirm the effectiveness and superiority of CFSum.

The success of Large Language Models (LLMs) has inspired the development of Multimodal Large Language Models (MLLMs) for unified understanding of vision and language. However, the increasing model size and computational complexity of large-scale MLLMs (l-MLLMs) limit their use in resource-constrained scenarios. Although small-scale MLLMs (s-MLLMs) are designed to reduce computational costs, they typically suffer from performance degradation. To mitigate this limitation, we propose a novel LLaVA-KD framework to transfer knowledge from l-MLLMs to s-MLLMs. Specifically, we introduce Multimodal Distillation (MDist) to transfer teacher model's robust representations across both visual and linguistic modalities, and Relation Distillation (RDist) to transfer teacher model's ability to capture visual token relationships. Additionally, we propose a three-stage training scheme to fully exploit the potential of the proposed distillation strategy: 1) Distilled Pre-Training to strengthen the alignment between visual-linguistic representations in s-MLLMs, 2) Supervised Fine-Tuning to equip the s-MLLMs with multimodal understanding capacity, and 3) Distilled Fine-Tuning to refine s-MLLM's knowledge. Our approach significantly improves s-MLLMs performance without altering the model architecture. Extensive experiments and ablation studies validate the effectiveness of each proposed component. Code will be available at https://github.com/Fantasyele/LLaVA-KD.

This paper presented DriveArena, the first high-fidelity closed-loop simulation system designed for driving agents navigating in real scenarios. DriveArena features a flexible, modular architecture, allowing for the seamless interchange of its core components: Traffic Manager, a traffic simulator capable of generating realistic traffic flow on any worldwide street map, and World Dreamer, a high-fidelity conditional generative model with infinite autoregression. This powerful synergy empowers any driving agent capable of processing real-world images to navigate in DriveArena's simulated environment. The agent perceives its surroundings through images generated by World Dreamer and output trajectories. These trajectories are fed into Traffic Manager, achieving realistic interactions with other vehicles and producing a new scene layout. Finally, the latest scene layout is relayed back into World Dreamer, perpetuating the simulation cycle. This iterative process fosters closed-loop exploration within a highly realistic environment, providing a valuable platform for developing and evaluating driving agents across diverse and challenging scenarios. DriveArena signifies a substantial leap forward in leveraging generative image data for the driving simulation platform, opening insights for closed-loop autonomous driving. Code will be available soon on GitHub: https://github.com/PJLab-ADG/DriveArena

We present Timer-XL, a causal Transformer for unified time series forecasting. To uniformly predict multidimensional time series, we generalize next token prediction, predominantly adopted for 1D token sequences, to multivariate next token prediction. The paradigm formulates various forecasting tasks as a long-context prediction problem. We opt for decoder-only Transformers that capture causal dependencies from varying-length contexts for unified forecasting, making predictions on non-stationary univariate time series, multivariate series with complicated dynamics and correlations, as well as covariate-informed contexts that include exogenous variables. Technically, we propose a universal TimeAttention to capture fine-grained intra- and inter-series dependencies of flattened time series tokens (patches), which is further enhanced by deft position embedding for temporal causality and variable equivalence. Timer-XL achieves state-of-the-art performance across task-specific forecasting benchmarks through a unified approach. Based on large-scale pre-training, Timer-XL achieves state-of-the-art zero-shot performance, making it a promising architecture for pre-trained time series models. Code is available at this repository: https://github.com/thuml/Timer-XL.

We introduce Sundial, a family of native, flexible, and scalable time series foundation models. To predict the next-patch's distribution, we propose a TimeFlow Loss based on flow-matching, which facilitates native pre-training of Transformers on continuous-valued time series without discrete tokenization. Conditioned on arbitrary-length time series, our models are pre-trained without specifying any prior distribution and can generate multiple probable predictions, achieving more flexibility in representation learning than using parametric densities. Towards time series foundation models, we leverage minimal but crucial adaptations of Transformers and curate TimeBench with one trillion time points, comprising mostly real-world datasets and synthetic data. By mitigating mode collapse via TimeFlow Loss, we pre-train a family of Sundial models on TimeBench, which achieve unprecedented model capacity and generalization performance. In addition to excellent scalability, Sundial achieves state-of-the-art results on both point and probabilistic forecasting benchmarks with a just-in-time inference speed, i.e., making zero-shot predictions within a few milliseconds. We believe that Sundial's pioneering generative forecasting capability can improve model reliability in real-world decision-making. Code is available at: https://github.com/thuml/Sundial.

Integrating natural language instructions and visual perception with decision-making is a critical challenge for embodied agents. Existing methods often struggle to balance the conciseness of language commands with the richness of video content. To bridge the gap between modalities, we propose extracting key spatiotemporal patterns from video that capture visual saliency and temporal evolution, referred to as dynamic representation. Building on this, we introduce DynaMind, a framework that enhances decision-making through dynamic reasoning. Specifically, we design an adaptive FrameScorer to evaluate video frames based on semantic consistency and visual saliency, assigning each frame an importance score. These scores are used to filter redundant video content and synthesize compact dynamic representations. Leveraging these representations, we predict critical future dynamics and apply a dynamic-guided policy to generate coherent and context-aware actions. Extensive results demonstrate that DynaMind significantly outperforms the baselines across several simulation benchmarks and real-world scenarios.

In this paper, we present a real-time photo-realistic SLAM method based on marrying Gaussian Splatting with LiDAR-Inertial-Camera SLAM. Most existing radiance-field-based SLAM systems mainly focus on bounded indoor environments, equipped with RGB-D or RGB sensors. However, they are prone to decline when expanding to unbounded scenes or encountering adverse conditions, such as violent motions and changing illumination. In contrast, oriented to general scenarios, our approach additionally tightly fuses LiDAR, IMU, and camera for robust pose estimation and photo-realistic online mapping. To compensate for regions unobserved by the LiDAR, we propose to integrate both the triangulated visual points from images and LiDAR points for initializing 3D Gaussians. In addition, the modeling of the sky and varying camera exposure have been realized for high-quality rendering. Notably, we implement our system purely with C++ and CUDA, and meticulously design a series of strategies to accelerate the online optimization of the Gaussian-based scene representation. Extensive experiments demonstrate that our method outperforms its counterparts while maintaining real-time capability. Impressively, regarding photo-realistic mapping, our method with our estimated poses even surpasses all the compared approaches that utilize privileged ground-truth poses for mapping. Our code will be released on project page https://xingxingzuo.github.io/gaussian_lic.

Realistic 3D object and scene reconstruction is pivotal in advancing fields such as world model simulation and embodied intelligence. In this paper, we introduce Hash-GS, a storage-efficient method for large-scale scene reconstruction using anchor-based 3D Gaussian Splatting (3DGS). The vanilla 3DGS struggles with high memory demands due to the large number of primitives, especially in complex or extensive scenes. Hash-GS addresses these challenges with a compact representation by leveraging high-dimensional features to parameterize primitive properties, stored in compact hash tables, which reduces memory usage while preserving rendering quality. It also incorporates adaptive anchor management to efficiently control the number of anchors and neural Gaussians. Additionally, we introduce an analytic 3D smoothing filter to mitigate aliasing and support Level-of-Detail for optimized rendering across varying intrinsic parameters. Experimental results on several datasets demonstrate that Hash-GS improves storage efficiency while maintaining competitive rendering performance, especially in large-scale scenes.

High-level robotic manipulation tasks demand flexible 6-DoF grasp estimation to serve as a basic function. Previous approaches either directly generate grasps from point-cloud data, suffering from challenges with small objects and sensor noise, or infer 3D information from RGB images, which introduces expensive annotation requirements and discretization issues. Recent methods mitigate some challenges by retaining a 2D representation to estimate grasp keypoints and applying Perspective-n-Point (PnP) algorithms to compute 6-DoF poses. However, these methods are limited by their non-differentiable nature and reliance solely on 2D supervision, which hinders the full exploitation of rich 3D information. In this work, we present KGN-Pro, a novel grasping network that preserves the efficiency and fine-grained object grasping of previous KGNs while integrating direct 3D optimization through probabilistic PnP layers. KGN-Pro encodes paired RGB-D images to generate Keypoint Map, and further outputs a 2D confidence map to weight keypoint contributions during re-projection error minimization. By modeling the weighted sum of squared re-projection errors probabilistically, the network effectively transmits 3D supervision to its 2D keypoint predictions, enabling end-to-end learning. Experiments on both simulated and real-world platforms demonstrate that KGN-Pro outperforms existing methods in terms of grasp cover rate and success rate. Project website: https://waitderek.github.io/kgnpro.

Extrinsic calibration is essential for multi-sensor fusion, existing methods rely on structured targets or fully-excited data, limiting real-world applicability. Online calibration further suffers from weak excitation, leading to unreliable estimates. To address these limitations, we propose a reinforcement learning (RL)-based extrinsic calibration framework that formulates extrinsic calibration as a decisionmaking problem, directly optimizes SE(3) extrinsics to enhance odometry accuracy. Our approach leverages a probabilistic Bingham distribution to model 3D rotations, ensuring stable optimization while inherently retaining quaternion symmetry. A trajectory alignment reward mechanism enables robust calibration without structured targets by quantitatively evaluating estimated tightly-coupled trajectory against a reference trajectory. Additionally, an automated data selection module filters uninformative samples, significantly improving efficiency and scalability for large-scale datasets. Extensive experiments on UAVs, UGVs, and handheld platforms demonstrate that our method outperforms traditional optimization-based approaches, achieving high-precision calibration even under weak excitation conditions. Our framework simplifies deployment on diverse robotic platforms by eliminating the need for high-quality initial extrinsics and enabling calibration from routine operating data. The code is available at https://github.com/APRIL-ZJU/learn-to-calibrate.

Semantic Scene Completion (SSC) constitutes a pivotal element in autonomous driving perception systems, tasked with inferring the 3D semantic occupancy of a scene from sensory data. To improve accuracy, prior research has implemented various computationally demanding and memory-intensive 3D operations, imposing significant computational requirements on the platform during training and testing. This paper proposes L2COcc, a lightweight camera-centric SSC framework that also accommodates LiDAR inputs. With our proposed efficient voxel transformer (EVT) and cross-modal knowledge modules, including feature similarity distillation (FSD), TPV distillation (TPVD) and prediction alignment distillation (PAD), our method substantially reduce computational burden while maintaining high accuracy. The experimental evaluations demonstrate that our proposed method surpasses the current state-of-the-art vision-based SSC methods regarding accuracy on both the SemanticKITTI and SSCBench-KITTI-360 benchmarks, respectively. Additionally, our method is more lightweight, exhibiting a reduction in both memory consumption and inference time by over 23% compared to the current state-of-the-arts method. Code is available at our project page: https://studyingfufu.github.io/L2COcc/.

Mobile robots are increasingly required to navigate and interact within unknown and unstructured environments to meet human demands. Demand-driven navigation (DDN) enables robots to identify and locate objects based on implicit human intent, even when object locations are unknown. However, traditional data-driven DDN methods rely on pre-collected data for model training and decision-making, limiting their generalization capability in unseen scenarios. In this paper, we propose CogDDN, a VLM-based framework that emulates the human cognitive and learning mechanisms by integrating fast and slow thinking systems and selectively identifying key objects essential to fulfilling user demands. CogDDN identifies appropriate target objects by semantically aligning detected objects with the given instructions. Furthermore, it incorporates a dual-process decision-making module, comprising a Heuristic Process for rapid, efficient decisions and an Analytic Process that analyzes past errors, accumulates them in a knowledge base, and continuously improves performance. Chain of Thought (CoT) reasoning strengthens the decision-making process. Extensive closed-loop evaluations on the AI2Thor simulator with the ProcThor dataset show that CogDDN outperforms single-view camera-only methods by 15%, demonstrating significant improvements in navigation accuracy and adaptability. The project page is available at https://yuehaohuang.github.io/CogDDN/.

Multimodal Large Language Models (MLLMs) have demonstrated excellent performance in video understanding but suffer from degraded effectiveness when processing long videos due to fixed-length contexts and weaknesses in modeling long-term dependencies. Retrieval-Augmented Generation (RAG) technology can mitigate these limitations through dynamic knowledge expansion, but existing RAG schemes for video understanding employ fixed retrieval paradigms that use uniform structures regardless of input query difficulty. This introduces redundant computational overhead and latency (*e.g.*, complex graph traversal operations) for simple queries (*e.g.*, frame-level object recognition) while potentially causing critical information loss due to insufficient retrieval granularity for multi-hop reasoning. Such single-step retrieval mechanisms severely constrain the model's balance between resource efficiency and cognitive depth. To address this, we first propose a novel AdaVideoRAG framework for long-video understanding, which uses a lightweight intent classifier to dynamically and adaptively allocate appropriate retrieval schemes, ranging from the simplest to the most sophisticated, for different video understanding tasks based on query complexity. We introduce an Omni-Knowledge Indexing module to extract valuable information from multi-modal signals for context modeling and build corresponding databases, *i.e.*, a text base from clip captions, ASR, and OCR; a visual base; and a graph for deep semantic understanding. This enables hierarchical knowledge access, integration, and generation from naive retrieval to graph retrieval, achieving an optimal balance between resource consumption and video understanding capabilities. Finally, we construct the HiVU benchmark for deep understanding evaluation. Extensive experiments show that our framework enhances the overall efficiency and accuracy of Video-QA for long videos and can be seamlessly integrated with existing MLLMs via lightweight API calls, establishing a new paradigm for adaptive retrieval augmentation in video analysis.

Diffusion models are advancing autonomous driving by enabling realistic data synthesis, predictive end-to-end planning, and closed-loop simulation, with a primary focus on temporally consistent generation. However, large-scale 3D scene generation requiring spatial coherence remains underexplored. In this paper, we present X-Scene, a novel framework for large-scale driving scene generation that achieves geometric intricacy, appearance fidelity, and flexible controllability. Specifically, X-Scene supports multi-granular control, including low-level layout conditioning driven by user input or text for detailed scene composition, and high-level semantic guidance informed by user intent and LLM-enriched prompts for efficient customization. To enhance geometric and visual fidelity, we introduce a unified pipeline that sequentially generates 3D semantic occupancy and corresponding multi-view images and videos, ensuring alignment and temporal consistency across modalities. We further extend local regions into large-scale scenes via consistency-aware outpainting, which extrapolates occupancy and images from previously generated areas to maintain spatial and visual coherence. The resulting scenes are lifted into high-quality 3DGS representations, supporting diverse applications such as simulation and scene exploration. Extensive experiments demonstrate that X-Scene substantially advances controllability and fidelity in large-scale scene generation, empowering data generation and simulation for autonomous driving.

This work focuses on developing parameter-efficient and lightweight models for dense predictions while trading off parameters, FLOPs, and performance. Our goal is to set up the new frontier of the 5 M magnitude lightweight model on various downstream tasks. Inverted Residual Block (IRB) serves as the infrastructure for lightweight CNNs, but no counterparts have been recognized by attention-based design. Our work rethinks the lightweight infrastructure of efficient IRB and practical components in Transformer from a unified perspective, extending CNN-based IRB to attention-based models and abstracting a one-residual Meta Mobile Block (MMBlock) for lightweight model design. Following neat but effective design criterion, we deduce a modern Improved Inverted Residual Mobile Block (i$^{2}$2RMB) and improve a hierarchical Efficient MOdel (EMOv2) with no elaborate complex structures. Considering the imperceptible latency for mobile users when downloading models under 4 G/5 G bandwidth and ensuring model performance, we investigate the performance upper limit of lightweight models with a magnitude of 5 M. Extensive experiments on various vision recognition, dense prediction, and image generation tasks demonstrate the superiority of our EMOv2 over state-of-the-art methods, e.g., EMOv2-1 M/2M/5 M achieve 72.3, 75.8, and 79.4 Top-1 that surpass equal-order CNN-/Attention-based models significantly. At the same time, EMOv2-5 M equipped RetinaNet achieves 41.5 mAP for object detection tasks that surpasses the previous EMO-5 M by +2.6$\uparrow$↑ . When employing the more robust training recipe, our EMOv2-5M eventually achieves 82.9 Top-1 accuracy, which elevates the performance of 5M magnitude models to a new level.

LiDAR panoptic segmentation (LPS) performs semantic and instance segmentation for things (foreground objects) and stuff (background elements), essential for scene perception and remote sensing. While most existing methods separate these tasks using distinct branches (i.e., semantic and instance), recent approaches have unified LPS through a query-based paradigm. However, the distinct spatial distributions of foreground objects and background elements in large-scale outdoor scenes pose challenges. This paper presents DQFormer, a novel framework for unified LPS that employs a decoupled query workflow to adapt to the characteristics of things and stuff in outdoor scenes. It first utilizes a feature encoder to extract multi-scale voxel-wise, point-wise, and BEV features. Then, a decoupled query generator proposes informative queries by localizing things/stuff positions and fusing multi-level BEV embeddings. A query-oriented mask decoder uses masked cross-attention to decode segmentation masks, which are combined with query semantics to produce panoptic results. Extensive experiments on large-scale outdoor scenes, including the vehicular datasets nuScenes and SemanticKITTI, as well as the aerial point cloud dataset DALES, show that DQFormer outperforms superior methods by +1.8%, +0.9%, and +3.5% in panoptic quality (PQ), respectively. Code is available at https://github.com/yuyang-cloud/DQFormer.

Intelligent Transportation System (ITS) is a critical component of smart cities, however, certain issues significantly limit the construction of ITS. On the one hand, as the core resource of ITS, traffic data often suffers from missing values due to sensor failure, communication interruption, and so on. On the other hand, traffic flow change is a complex dynamic process that resulted from periodic changes caused by social activities, which increases the difficulty of data completion. To address these issues, the Message Passing Period-Aware Imputation Network (MPPAIN) is proposed. Firstly, the spatial-temporal information is transmitted sequentially in time order by the Message Passing Block based on gated recurrent unit, and the missing values are preliminarily estimated. Then, the output is fed to the Period-Aware Block to find the main frequency components that represent the changes of the traffic flow in the frequency domain through the Fourier transform. Subsequently, the traffic flow is divided according to the major periods, and the second time estimation is completed by extracting the intra-periodic and inter-periodic features simultaneously through the convolutional neural network. Finally, a bi-directional structure is designed by reversing the input traffic flow in time order to further extract the spatial-temporal and periodic features from the future to the past. Experiments demonstrate that the proposed model has excellent data imputation capabilities in various simulated missing rates and missing scenarios on several real traffic datasets.

Binary neural network (BNN) converts full-precision weights and activations into their extreme 1-bit counterparts, making it particularly suitable for deployment on lightweight mobile devices. While BNNs are typically formulated as a constrained optimization problem and optimized in the binarized space, general neural networks are formulated as an unconstrained optimization problem and optimized in the continuous space. This article introduces the hyperbolic BNN (HBNN) by leveraging the framework of hyperbolic geometry to optimize the constrained problem. Specifically, we transform the constrained problem in hyperbolic space into an unconstrained one in Euclidean space using the Riemannian exponential map. On the other hand, we also propose the exponential parametrization cluster (EPC) method, which, compared with the Riemannian exponential map, shrinks the segment domain based on a diffeomorphism. This approach increases the probability of weight flips, thereby maximizing the information gain in BNNs. Experimental results on CIFAR10, CIFAR100, and ImageNet classification datasets with VGGsmall, ResNet18, and ResNet34 models illustrate the superior performance of our HBNN over state-of-the-art methods.

Model compression methods are being developed to bridge the gap between the massive scale of neural networks and the limited hardware resources on edge devices. Since most real-world applications deployed on resource-limited hardware platforms typically have multiple hardware constraints simultaneously, most existing model compression approaches that only consider optimizing one single hardware objective are ineffective. In this article, we propose an automated pruning method called multi-constrained model compression (MCMC) that allows for the optimization of multiple hardware targets, such as latency, floating point operations (FLOPs), and memory usage, while minimizing the impact on accuracy. Specifically, we propose an improved multi-objective reinforcement learning (MORL) algorithm, the one-stage envelope deep deterministic policy gradient (DDPG) algorithm, to determine the pruning strategy for neural networks. Our improved one-stage envelope DDPG algorithm reduces exploration time and offers greater flexibility in adjusting target priorities, enhancing its suitability for pruning tasks. For instance, on the visual geometry group (VGG)-16 network, our method achieved an 80% reduction in FLOPs, a reduction in memory usage, and a acceleration, with an accuracy improvement of 0.09% compared with the baseline. For larger datasets, such as ImageNet, we reduced FLOPs by 50% for MobileNet-V1, resulting in a faster speed and memory compression, while maintaining the same accuracy. When applied to edge devices, such as JETSON XAVIER NX, our method resulted in a 71% reduction in FLOPs for MobileNet-V1, leading to a faster speed, memory compression, and an accuracy improvement.

Filter pruning is a significant feature selection technique to shrink the existing feature fusion schemes (especially on convolution calculation and model size), which helps to develop more efficient feature fusion models while maintaining state-of-the-art performance. In addition, it reduces the storage and computation requirements of deep neural networks (DNNs) and accelerates the inference process dramatically. Existing methods mainly rely on manual constraints such as normalization to select the filters. A typical pipeline comprises two stages: first pruning the original neural network and then fine-tuning the pruned model. However, choosing a manual criterion can be somehow tricky and stochastic. Moreover, directly regularizing and modifying filters in the pipeline suffer from being sensitive to the choice of hyperparameters, thus making the pruning procedure less robust. To address these challenges, we propose to handle the filter pruning issue through one stage: using an attention-based architecture that adaptively fuses the filter selection with filter learning in a unified network. Specifically, we present a pruning method named adding before pruning (ABP) to make the model focus on the filters of higher significance by training instead of man-made criteria such as norm, rank, etc. First, we add an auxiliary attention layer into the original model and set the significance scores in this layer to be binary. Furthermore, to propagate the gradients in the auxiliary attention layer, we design a specific gradient estimator and prove its effectiveness for convergence in the graph flow through mathematical derivation. In the end, to relieve the dependence on the complicated prior knowledge for designing the thresholding criterion, we simultaneously prune and train the filters to automatically eliminate network redundancy with recoverability. Extensive experimental results on the two typical image classification benchmarks, CIFAR-10 and ILSVRC-2012, illustrate that the proposed approach performs favorably against previous state-of-the-art filter pruning algorithms.

The canonical approach to video action recognition dictates a neural network model to do a classic and standard 1-of-N majority vote task. They are trained to predict a fixed set of predefined categories, limiting their transferability on new datasets with unseen concepts. In this article, we provide a new perspective on action recognition by attaching importance to the semantic information of label texts rather than simply mapping them into numbers. Specifically, we model this task as a video-text matching problem within a multimodal learning framework, which strengthens the video representation with more semantic language supervision and enables our model to do zero-shot action recognition without any further labeled data or parameters' requirements. Moreover, to handle the deficiency of label texts and make use of tremendous web data, we propose a new paradigm based on this multimodal learning framework for action recognition, which we dub "pre-train, adapt and fine-tune." This paradigm first learns powerful representations from pre-training on a large amount of web image-text or video-text data. Then, it makes the action recognition task to act more like pre-training problems via adaptation engineering. Finally, it is fine-tuned end-to-end on target datasets to obtain strong performance. We give an instantiation of the new paradigm, ActionCLIP, which not only has superior and flexible zero-shot/few-shot transfer ability but also reaches a top performance on general action recognition task, achieving 83.8% top-1 accuracy on Kinetics-400 with a ViT-B/16 as the backbone. Code is available at https://github.com/sallymmx/ActionCLIP.git.

The point cloud-based 3D single object tracking plays an indispensable role in autonomous driving. However, the application of 3D object tracking in the real world is still challenging due to the inherent sparsity and self-occlusion of point cloud data. Therefore, it is necessary to exploit as much useful information from limited data as we can. Since 3D object tracking is a video-level task, the appearance of objects changes gradually over time, and there is rich spatiotemporal contextual information among historical frames. However, existing methods do not fully utilize this information. To address this, we propose a new method called SCTrack, which utilizes a memory-based paradigm to exploit spatiotemporal contextual information. SCTrack incorporates both long-term and short-term memory banks to store the spatiotemporal features of targets from historical frames. By doing so, the tracker can benefit from the entire video sequence and make more informed predictions. Additionally, SCTrack extracts the mask prior to augmenting the target representation, improving the target-background discriminability. Extensive experiments on KITTI, nuScenes, and Waymo Open datasets verify the effectiveness of our proposed method.

Recently, the rise of large-scale vision-language pretrained models like CLIP, coupled with the technology of Parameter-Efficient FineTuning (PEFT), has captured substantial attraction in video action recognition. Nevertheless, prevailing approaches tend to prioritize strong supervised performance at the expense of compromising the models' generalization capabilities during transfer. In this paper, we introduce a novel Multimodal, Multi-task CLIP adapting framework named M2-CLIP to address these challenges, preserving both high supervised performance and robust transferability. Firstly, to enhance the individual modality architectures, we introduce multimodal adapters to both the visual and text branches. Specifically, we design a novel visual TED-Adapter, that performs global Temporal Enhancement and local temporal Difference modeling to improve the temporal representation capabilities of the visual encoder. Moreover, we adopt text encoder adapters to strengthen the learning of semantic label information. Secondly, we design a multi-task decoder with a rich set of supervisory signals, including the original contrastive learning head, a cross-modal classification head, a cross-modal masked language modeling head, and a visual classification head. This multi-task decoder adeptly satisfies the need for strong supervised performance within a multimodal framework. Experimental results validate the efficacy of our approach, demonstrating exceptional performance in supervised learning while maintaining strong generalization in zero-shot scenarios.

While large models have achieved significant progress in computer vision, challenges such as optimization complexity, the intricacy of transformer architectures, computational constraints, and practical application demands highlight the importance of simpler model designs in medical image segmentation. This need is particularly pronounced in mobile medical devices, which require lightweight, deployable models with real-time performance. However, existing lightweight models often suffer from poor robustness across datasets, limiting their widespread adoption. To address these challenges, this paper introduces LV-UNet, a lightweight and vanilla model that leverages pre-trained MobileNetv3-Large backbones and incorporates fusible modules. LV-UNet employs an enhanced deep training strategy and switches to a deployment mode during inference by re-parametrization, significantly reducing parameter count and computational overhead. Experimental results on ISIC 2016, BUSI, CVC-ClinicDB, CVC-ColonDB, and Kvair-SEG datasets demonstrate a better trade-off between performance and the computational load. The code will be released at https://github.com/juntaoJianggavin/LV-UNet.

Multimodal Visual Object Tracking (VOT) has recently gained significant attention due to its robustness. Early research focused on fully fine-tuning RGB-based trackers, which was inefficient and lacked generalized representation due to the scarcity of multimodal data. Therefore, recent studies have utilized prompt tuning to transfer pre-trained RGB-based trackers to multimodal data. However, the modality gap limits pre-trained knowledge recall, and the dominance of the RGB modality persists, preventing the full utilization of information from other modalities. To address these issues, we propose a novel symmetric multimodal tracking framework called SDSTrack. We introduce lightweight adaptation for efficient fine-tuning, which directly transfers the feature extraction ability from RGB to other domains with a small number of trainable parameters and integrates multimodal features in a balanced, symmetric manner. Furthermore, we design a complementary masked patch distillation strategy to enhance the robustness of trackers in complex environments, such as extreme weather, poor imaging, and sensor failure. Extensive experiments demonstrate that SDSTrack outperforms state-of-the-art methods in various multimodal tracking scenarios, including RGB+Depth, RGB+Thermal, and RGB+Event tracking, and exhibits impressive results in extreme conditions. Our source code is available at: https://github.com/hoqolo/SDSTrack.

N:m sparsity has received increasing attention due to its remarkable performance and latency trade-off compared with structured and unstructured sparsity. How-ever, existing N:m sparsity methods do not differentiate the relative importance of weights among blocks and leave important weights underappreciated. Besides, they di-rectly apply N:m sparsity to the whole network, which will cause severe information loss. Thus, they are still sub-optimal. In this paper, we propose an efficient and effective Multi-Axis Query methodology, dubbed as MaxQ, to rectify these problems. During the training, MaxQ employs a dynamic approach to generate soft N:m masks, considering the weight importance across multiple axes. This method enhances the weights with more importance and ensures more effective updates. Meanwhile, a spar-sity strategy that gradually increases the percentage of N:m weight blocks is applied, which allows the network to heal from the pruning-induced damage progressively. During the runtime, the N:m soft masks can be precom-puted as constants and folded into weights without causing any distortion to the sparse pattern and incurring ad-ditional computational overhead. Comprehensive experi-ments demonstrate that MaxQ achieves consistent improve-ments across diverse CNN architectures in various com-puter vision tasks, including image classification, object detection and instance segmentation. For ResNet50 with 1:16 sparse pattern, MaxQ can achieve 74.6% top-1 ac-curacy on ImageNet and improve by over 2.8% over the state-of-the-art. Codes and checkpoints are available at https://github.com/JingyangXiang/MaxQ.

This paper presents Paint3D, a novel coarse-to-fine generative framework that is capable of producing high-resolution, lighting-less, and diverse 2K UV texture maps for untextured 3D meshes conditioned on text or image inputs. The key challenge addressed is generating high-quality textures without embedded illumination information, which allows the textures to be re-lighted or re-edited within modern graphics pipelines. To achieve this, our method first leverages a pre-trained depth-aware 2D diffusion model to generate view-conditional images and per-form multi-view texture fusion, producing an initial coarse texture map. However, as 2D models cannot fully repre-sent 3D shapes and disable lighting effects, the coarse texture map exhibits incomplete areas and illumination artifacts. To resolve this, we train separate UV Inpainting and UVHD diffusion models specialized for the shape-aware re-finement of incomplete areas and the removal of illumination artifacts. Through this coarse-to-fine process, Paint3D can produce high-quality 2K UV textures that maintain se-mantic consistency while being lighting-less, significantly advancing the state-of-the-art in texturing 3D objects.

The massive parameters and computational demands hinder the widespread application of Large Language Models (LLMs).Network pruning provides a practical solution to this problem.However, existing pruning works for LLMs mainly focus on unstructured pruning or necessitate post-pruning fine-tuning.The former relies on special hardware to accelerate computation, while the latter may need substantial computational resources.In this paper, we introduce a retraining-free structured pruning method called SoBP (Structured Optimal Brain Pruning).It leverages global first-order information to select pruning structures, then refines them with a local greedy approach, and finally adopts module-wise reconstruction to mitigate information loss.We assess the effectiveness of SoBP across 14 models from 3 LLM families on 8 distinct datasets.Experimental results demonstrate that SoBP outperforms current state-of-the-art methods.

Self-supervised contrastive learning (SSCL) has performed excellently on time series classification tasks. Most SSCL- based classification algorithms generate positive and negative samples in the time or frequency domains, focusing on mining similarities between them. However, two issues are not well addressed in the SSCL framework: the sampling bias and the task-agnostic representation problems. Sampling bias indicates fake negative sample selection in SSCL, and task- agnostic representation results in the unknown correlation between the extracted feature and downstream tasks. To address the issues, we propose Debiased Contrastive learning with weak Supervision framework, abbreviated as DCS. It employs the clustering operation to remove fake negative samples and introduces weak supervisory signals into the SSCL framework to guide feature extraction. Additionally, we propose a channel augmentation method that allows the DCS to extract features from local and global perspectives simultaneously. The comprehensive experiments on the widely used datasets show that DCS achieves performance superior to state-of-the-art methods on the widely used popular benchmark datasets.

Self-supervised learning (SSL) shows impressive performance in many tasks lacking sufficient labels. In this paper, we study SSL in time series anomaly detection (TSAD) by incorporating the characteristics of time series data. Specifically, we build an anomaly detection algorithm consisting of global pattern learning and local association learning. The global pattern learning module builds encoder and decoder to reconstruct the raw time series data to detect global anomalies. To complement the limitation of the global pattern learning that ignores local associations between anomaly points and their adjacent windows, we design a local association learning module, which leverages contrastive predictive coding (CPC) to transform the identification of anomaly points into positive pairs identification. Motivated by the observation that adjusting the distance between the history window and the time point to be detected directly impacts the detection performance in the CPC framework, we further propose a skip-step CPC scheme in the local association learning module which adjusts the distance for better construction of the positive pairs and detection results. The experimental results show that the proposed algorithm achieves superior performance on SMD and PSM datasets in comparison with 12 state-of-the-art algorithms.

The conjugate gradient method is a crucial first-order optimization method that generally converges faster than the steepest descent method, and its computational cost is much lower than that of second-order methods. However, while various types of conjugate gradient methods have been studied in Euclidean spaces and on Riemannian manifolds, there is little study for those in distributed scenarios. This paper proposes a decentralized Riemannian conjugate gradient descent (DRCGD) method that aims at minimizing a global function over the Stiefel manifold. The optimization problem is distributed among a network of agents, where each agent is associated with a local function, and the communication between agents occurs over an undirected connected graph. Since the Stiefel manifold is a non-convex set, a global function is represented as a finite sum of possibly non-convex (but smooth) local functions. The proposed method is free from expensive Riemannian geometric operations such as retractions, exponential maps, and vector transports, thereby reducing the computational complexity required by each agent. To the best of our knowledge, DRCGD is the first decentralized Riemannian conjugate gradient algorithm to achieve global convergence over the Stiefel manifold.

The recent boom of linear forecasting models questions the ongoing passion for architectural modifications of Transformer-based forecasters. These forecasters leverage Transformers to model the global dependencies over temporal tokens of time series, with each token formed by multiple variates of the same timestamp. However, Transformers are challenged in forecasting series with larger lookback windows due to performance degradation and computation explosion. Besides, the embedding for each temporal token fuses multiple variates that represent potential delayed events and distinct physical measurements, which may fail in learning variate-centric representations and result in meaningless attention maps. In this work, we reflect on the competent duties of Transformer components and repurpose the Transformer architecture without any modification to the basic components. We propose iTransformer that simply applies the attention and feed-forward network on the inverted dimensions. Specifically, the time points of individual series are embedded into variate tokens which are utilized by the attention mechanism to capture multivariate correlations; meanwhile, the feed-forward network is applied for each variate token to learn nonlinear representations. The iTransformer model achieves state-of-the-art on challenging real-world datasets, which further empowers the Transformer family with promoted performance, generalization ability across different variates, and better utilization of arbitrary lookback windows, making it a nice alternative as the fundamental backbone of time series forecasting. Code is available at this repository: https://github.com/thuml/iTransformer.

Deep learning has contributed remarkably to the advancement of time series analysis. Still, deep models can encounter performance bottlenecks in real-world data-scarce scenarios, which can be concealed due to the performance saturation with small models on current benchmarks. Meanwhile, large models have demonstrated great powers in these scenarios through large-scale pre-training. Continuous progress has been achieved with the emergence of large language models, exhibiting unprecedented abilities such as few-shot generalization, scalability, and task generality, which are however absent in small deep models. To change the status quo of training scenario-specific small models from scratch, this paper aims at the early development of large time series models (LTSM). During pre-training, we curate large-scale datasets with up to 1 billion time points, unify heterogeneous time series into single-series sequence (S3) format, and develop the GPT-style architecture toward LTSMs. To meet diverse application needs, we convert forecasting, imputation, and anomaly detection of time series into a unified generative task. The outcome of this study is a Time Series Transformer (Timer), which is generative pre-trained by next token prediction and adapted to various downstream tasks with promising capabilities as an LTSM. Code and datasets are available at: https://github.com/thuml/Large-Time-Series-Model.

Place recognition is a challenging but crucial task in robotics. Current description-based methods may be limited by representation capabilities, while pairwise similarity-based methods require exhaustive searches, which is time-consuming. In this paper, we present a novel coarse-to-fine approach to address these problems, which combines BEV (Bird’s Eye View) feature extraction, coarse-grained matching and fine-grained verification. In the coarse stage, our approach utilizes an attention-guided network to generate attention-guided descriptors. We then employ a fast affinity-based candidate selection process to identify the Top-K most similar candidates. In the fine stage, we estimate pairwise overlap among the narrowed-down place candidates to determine the final match. Experimental results on the KITTI and KITTI-360 datasets demonstrate that our approach outperforms state-of-the-art methods. The code will be released publicly soon.

We present a novel approach for metric dense depth estimation based on the fusion of a single-view image and a sparse, noisy Radar point cloud. The direct fusion of heterogeneous Radar and image data, or their encodings, tends to yield dense depth maps with significant artifacts, blurred boundaries, and suboptimal accuracy. To circumvent this issue, we learn to augment versatile and robust monocular depth prediction with the dense metric scale induced from sparse and noisy Radar data. We propose a Radar-Camera framework for highly accurate and fine-detailed dense depth estimation with four stages, including monocular depth prediction, global scale alignment of monocular depth with sparse Radar points, quasi-dense scale estimation through learning the association between Radar points and image patches, and local scale refinement of dense depth using a scale map learner. Our proposed method significantly outperforms the state-of-the-art Radar-Camera depth estimation methods by reducing the mean absolute error (MAE) of depth estimation by 25.6% and 40.2% on the challenging nuScenes dataset and our self-collected ZJU-4DRadarCam dataset, respectively. Our code and dataset will be released at https://github.com/MMOCKING/RadarCam-Depth.

Tracking a specific person in 3D scene is gaining momentum due to its numerous applications in robotics. Currently, most 3D trackers focus on driving scenarios with neglected jitter and uncomplicated surroundings, which results in their severe degeneration in complex environments, especially on jolting robot platforms (only 20-60% success rate). To improve the accuracy, a Point-Video-based Transformer Tracking model (PVTrack) is presented for robots. It is the first multi-modal 3D human tracking work that incorporates point clouds together with RGB videos to achieve information complementarity. Moreover, PVTrack proposes the Siamese Point-Video Transformer for feature aggregation to overcome dynamic environments, which captures more target-aware information through the hierarchical attention mechanism adaptively. Considering the violent shaking on robots and rugged terrains, a lateral Human-ware Proposal Network is designed together with an Anti-shake Proposal Compensation module. It alleviates the disturbance caused by complex scenes as well as the particularity of the robot platform. Experiments show that our method achieves state-of-the-art performance on both KITTI/Waymo datasets and a quadruped robot for various indoor and outdoor scenes.

Visual bird’s eye view (BEV) semantic segmentation helps autonomous vehicles understand the surrounding environment only from front-view (FV) images, including static elements (e.g., roads) and dynamic elements (e.g., vehicles, pedestrians). However, the high cost of annotation procedures of full-supervised methods limits the capability of the visual BEV semantic segmentation, which usually needs HD maps, 3D object bounding boxes, and camera extrinsic matrixes. In this paper, we present a novel semi-supervised framework for visual BEV semantic segmentation to boost performance by exploiting unlabeled images during the training. A consistency loss that makes full use of unlabeled data is then proposed to constrain the model on not only semantic prediction but also the BEV feature. Furthermore, we propose a novel and effective data augmentation method named conjoint rotation which reasonably augments the dataset while maintaining the geometric relationship between the FV images and the BEV semantic segmentation. Extensive experiments on the nuScenes dataset show that our semi-supervised framework can effectively improve prediction accuracy. To the best of our knowledge, this is the first work that explores improving visual BEV semantic segmentation performance using unlabeled data. The code is available at https://github.com/Junyu-Z/Semi-BEVseg.

Event cameras have garnered considerable attention due to their advantages over traditional cameras in low power consumption, high dynamic range, and no motion blur. This paper proposes a monocular event-inertial odometry incorporating an adaptive decay kernel-based time surface with polarity-aware tracking. We utilize an adaptive decay-based Time Surface to extract texture information from asynchronous events, which adapts to the dynamic characteristics of the event stream and enhances the representation of environmental textures. However, polarity-weighted time surfaces suffer from event polarity shifts during changes in motion direction. To mitigate its adverse effects on feature tracking, we optimize the feature tracking by incorporating an additional polarityinverted time surface to enhance the robustness. Comparative analysis with visual-inertial and event-inertial odometry methods shows that our approach outperforms state-of-the-art techniques, with competitive results across various datasets.

Cooperative path planning, a crucial aspect of multi-agent systems research, serves a variety of sectors, including military, agriculture, and industry. Many existing algorithms, however, come with certain limitations, such as simplified kinematic models and inadequate support for multiple group scenarios. Focusing on the planning problem associated with a nonholonomic Ackermann model for Unmanned Ground Vehicles (UGV), we propose a leaderless, hierarchical Search-Based Cooperative Motion Planning (SCMP) method. The high-level utilizes a binary conflict search tree to minimize runtime, while the low-level fabricates kinematically feasible, collision-free paths that are shape-constrained. Our algorithm can adapt to scenarios featuring multiple groups with different shapes, outlier agents, and elaborate obstacles. We conduct algorithm comparisons, performance testing, simulation, and real-world testing, verifying the effectiveness and applicability of our algorithm. The implementation of our method will be open-sourced at https://github.com/WYCUniverStar/SCMP.

The goal of this paper is to strike a feasible tracking paradigm that can make 3D human trackers applicable on robot platforms and enable more high-level tasks. Till now, two fundamental problems haven’t been adequately addressed. One is the computational cost lightweight enough for robotic deployment, and the other is the easily-influenced accuracy varied greatly in complex real environments. In this paper, a robotic-centric tracking paradigm called MATNet is proposed that directly matches the LiDAR point clouds and RGB videos through end-to-end learning. To improve the low accuracy of human tracking against disturbance, a coarse-to-fine Transformer along with target-ware augmentation is proposed by fusing RGB videos and point clouds through a pyramid encoding and decoding strategy. To better meet the real-time requirement of actual robot deployment, we introduce the parameter-efficient adaptation tuning that greatly shortens the model’s training time. Furthermore, we also propose a five-step Anti-shake Refinement strategy and have added human prior values to overcome the strong shaking on the robot plat-form. Extensive experiments confirm that MATNet significantly outperforms the previous state-of-the-art on both open-source datasets and large-scale robotic datasets.

Deep reinforcement learning controllers with exteroception have enabled quadrupedal robots to traverse terrain robustly. However, most of these controllers heavily depend on complex reward functions and suffer from poor convergence. This work proposes a novel learning framework called derived-action optimization. The derived action is defined as a high-level representation of a policy and can be introduced into the reward function to guide decision-making behaviors. The proposed derived-action optimization method is applied to learn safer quadrupedal locomotion, achieving fast convergence and better performance. Specifically, we choose the foothold as the derived action and optimize the flatness of the terrain around the foothold to reduce potential sliding and collisions. Extensive experiments demonstrate the high safety and effectiveness of our method.

Co-speech gesture generation is crucial for producing synchronized and realistic human gestures that accompany speech, enhancing the animation of lifelike avatars in virtual environments. While diffusion models have shown impressive capabilities, current approaches often overlook a wide range of modalities and their interactions, resulting in less dynamic and contextually varied gestures. To address these challenges, we present MambaGesture, a novel framework integrating a Mamba-based attention block, MambaAttn, with a multi-modality feature fusion module, SEAD. The MambaAttn block combines the sequential data processing strengths of the Mamba model with the contextual richness of attention mechanisms, enhancing the temporal coherence of generated gestures. SEAD adeptly fuses audio, text, style, and emotion modalities, employing disentanglement to deepen the fusion process and yield gestures with greater realism and diversity. Our approach, rigorously evaluated on the multi-modal BEAT dataset, demonstrates significant improvements in Fréchet Gesture Distance (FGD), diversity scores, and beat alignment, achieving state-of-the-art performance in co-speech gesture generation.

Text-guided diffusion models have significantly advanced image editing, enabling high-quality and diverse modifications driven by text prompts. However, effective editing requires inverting the source image into a latent space, a process often hindered by prediction errors inherent in DDIM inversion. These errors accumulate during the diffusion process, resulting in inferior content preservation and edit fidelity, especially with conditional inputs. We address these challenges by investigating the primary contributors to error accumulation in DDIM inversion and identify the singularity problem in traditional noise schedules as a key issue. To resolve this, we introduce the *Logistic Schedule*, a novel noise schedule designed to eliminate singularities, improve inversion stability, and provide a better noise space for image editing. This schedule reduces noise prediction errors, enabling more faithful editing that preserves the original content of the source image. Our approach requires no additional retraining and is compatible with various existing editing methods. Experiments across eight editing tasks demonstrate the Logistic Schedule's superior performance in content preservation and edit fidelity compared to traditional noise schedules, highlighting its adaptability and effectiveness. The project page is available at https://lonelvino.github.io/SYE/.

Foundation models of time series have not been fully developed due to the limited availability of time series corpora and the underexploration of scalable pre-training. Based on the similar sequential formulation of time series and natural language, increasing research demonstrates the feasibility of leveraging large language models (LLM) for time series. Nevertheless, the inherent autoregressive property and decoder-only architecture of LLMs have not been fully considered, resulting in insufficient utilization of LLM abilities. To fully revitalize the general-purpose token transition and multi-step generation capability of large language models, we propose AutoTimes to repurpose LLMs as autoregressive time series forecasters, which projects time series into the embedding space of language tokens and autoregressively generates future predictions with arbitrary lengths. Compatible with any decoder-only LLMs, the consequent forecaster exhibits the flexibility of the lookback length and scalability with larger LLMs. Further, we formulate time series as prompts, extending the context for prediction beyond the lookback window, termed in-context forecasting. By introducing LLM-embedded textual timestamps, AutoTimes can utilize chronological information to align multivariate time series. Empirically, AutoTimes achieves state-of-the-art with 0.1% trainable parameters and over $5\times$ training/inference speedup compared to advanced LLM-based forecasters. Code is available at this repository: https://github.com/thuml/AutoTimes.

Autonomous driving has advanced significantly due to sensors, machine learning, and artificial intelligence improvements. However, prevailing methods struggle with intricate scenarios and causal relationships, hindering adaptability and interpretability in varied environments. To address the above problems, we introduce LeapAD, a novel paradigm for autonomous driving inspired by the human cognitive process. Specifically, LeapAD emulates human attention by selecting critical objects relevant to driving decisions, simplifying environmental interpretation, and mitigating decision-making complexities. Additionally, LeapAD incorporates an innovative dual-process decision-making module, which consists of an Analytic Process (System-II) for thorough analysis and reasoning, along with a Heuristic Process (System-I) for swift and empirical processing. The Analytic Process leverages its logical reasoning to accumulate linguistic driving experience, which is then transferred to the Heuristic Process by supervised fine-tuning. Through reflection mechanisms and a growing memory bank, LeapAD continuously improves itself from past mistakes in a closed-loop environment. Closed-loop testing in CARLA shows that LeapAD outperforms all methods relying solely on camera input, requiring 1-2 orders of magnitude less labeled data. Experiments also demonstrate that as the memory bank expands, the Heuristic Process with only 1.8B parameters can inherit the knowledge from a GPT-4 powered Analytic Process and achieve continuous performance improvement. Project page: https://pjlab-adg.github.io/LeapAD

Deep models have demonstrated remarkable performance in time series forecasting. However, due to the partially-observed nature of real-world applications, solely focusing on the target of interest, so-called endogenous variables, is usually insufficient to guarantee accurate forecasting. Notably, a system is often recorded into multiple variables, where the exogenous variables can provide valuable external information for endogenous variables. Thus, unlike well-established multivariate or univariate forecasting paradigms that either treat all the variables equally or ignore exogenous information, this paper focuses on a more practical setting: time series forecasting with exogenous variables. We propose a novel approach, TimeXer, to ingest external information to enhance the forecasting of endogenous variables. With deftly designed embedding layers, TimeXer empowers the canonical Transformer with the ability to reconcile endogenous and exogenous information, where patch-wise self-attention and variate-wise cross-attention are used simultaneously. Moreover, global endogenous tokens are learned to effectively bridge the causal information underlying exogenous series into endogenous temporal patches. Experimentally, TimeXer achieves consistent state-of-the-art performance on twelve real-world forecasting benchmarks and exhibits notable generality and scalability. Code is available at this repository: https://github.com/thuml/TimeXer.

In this paper, we study Discretized Neural Networks (DNNs) composed of low-precision weights and activations, which suffer from either infinite or zero gradients due to the non-differentiable discrete function during training. Most training-based DNNs in such scenarios employ the standard Straight-Through Estimator (STE) to approximate the gradient w.r.t. discrete values. However, the use of STE introduces the problem of gradient mismatch, arising from perturbations in the approximated gradient. To address this problem, this paper reveals that this mismatch can be interpreted as a metric perturbation in a Riemannian manifold, viewed through the lens of duality theory. Building on information geometry, we construct the Linearly Nearly Euclidean (LNE) manifold for DNNs, providing a background for addressing perturbations. By introducing a partial differential equation on metrics, i.e., the Ricci flow, we establish the dynamical stability and convergence of the LNE metric with the $L^2$-norm perturbation. In contrast to previous perturbation theories with convergence rates in fractional powers, the metric perturbation under the Ricci flow exhibits exponential decay in the LNE manifold. Experimental results across various datasets demonstrate that our method achieves superior and more stable performance for DNNs compared to other representative training-based methods.

Few-Shot Instance Segmentation (FSIS) requires detecting and segmenting novel classes with limited support examples. Existing methods based on Region Proposal Networks (RPNs) face two issues: 1) overfitting suppresses novel class objects and 2) dual-branch models require complex spatial correlation strategies to prevent spatial information loss when generating class prototypes. We introduce a unified framework, Reference Twice (RefT), to exploit the relationship between support and query features for FSIS and related tasks. Our three main contributions are: 1) a novel transformer-based baseline that avoids overfitting, offering a new direction for FSIS; 2) demonstrating that support object queries encode key factors after base training, allowing query features to be enhanced twice at both feature and query levels using simple cross-attention, thus avoiding complex spatial correlation interaction; and 3) introducing a class-enhanced base knowledge distillation loss to address the issue of DETR-like models struggling with incremental settings due to the input projection layer, enabling easy extension to incremental FSIS. Extensive experimental evaluations on the COCO dataset under three FSIS settings demonstrate that our method performs favorably against existing approaches across different shots, e.g., +8.2/+9.4 performance gain over state-of-the-art methods with 10/30-shots.

Cross-domain generalizable depth estimation aims to estimate the depth of target domains (i.e., real-world) using models trained on the source domains (i.e., synthetic). Previous methods mainly use additional real-world domain datasets to extract depth specific information for cross-domain generalizable depth estimation. Unfortunately, due to the large domain gap, adequate depth specific information is hard to obtain and interference is difficult to remove, which limits the performance. To relieve these problems, we propose a domain generalizable feature extraction network with adaptive guidance fusion (AGDF-Net) to fully acquire essential features for depth estimation at multi-scale feature levels. Specifically, our AGDF-Net first separates the image into initial depth and weak-related depth components with reconstruction and contrary losses. Subsequently, an adaptive guidance fusion module is designed to sufficiently intensify the initial depth features for domain generalizable intensified depth features acquisition. Finally, taking intensified depth features as input, an arbitrary depth estimation network can be used for real-world depth estimation. Using only synthetic datasets, our AGDF-Net can be applied to various real-world datasets (i.e., KITTI, NYUDv2, NuScenes, DrivingStereo and CityScapes) with state-of-the-art performances. Furthermore, experiments with a small amount of real-world data in a semi-supervised setting also demonstrate the superiority of AGDF-Net over state-of-the-art approaches.

Self-supervised learning (SSL) has recently achieved impressive performance on various time series tasks. The most prominent advantage of SSL is that it reduces the dependence on labeled data. Based on the pre-training and fine-tuning strategy, even a small amount of labeled data can achieve high performance. Compared with many published self-supervised surveys on computer vision and natural language processing, a comprehensive survey for time series SSL is still missing. To fill this gap, we review current state-of-the-art SSL methods for time series data in this article. To this end, we first comprehensively review existing surveys related to SSL and time series, and then provide a new taxonomy of existing time series SSL methods by summarizing them from three perspectives: generative-based, contrastive-based, and adversarial-based. These methods are further divided into ten subcategories with detailed reviews and discussions about their key intuitions, main frameworks, advantages and disadvantages. To facilitate the experiments and validation of time series SSL methods, we also summarize datasets commonly used in time series forecasting, classification, anomaly detection, and clustering tasks. Finally, we present the future directions of SSL for time series analysis.

This article investigates the multi-vehicle task assignment and motion planning (MVTAMP) problem. In a dense environment, a fleet of non-holonomic vehicles is appointed to visit a series of target positions and then move to a specific ending area for real-world applications such as clearing threat targets, aid rescue, and package delivery. We presented a novel hierarchical method to simultaneously address the multiple vehicles’ task assignment and motion planning problem. Unlike most related work, our method considers the MVTAMP problem applied to non-holonomic vehicles in large-scale scenarios. At the high level, we proposed a novel distributed algorithm to address task assignment, which produces a closer to the optimal task assignment scheme by reducing the intersection paths between vehicles and tasks or between tasks and tasks. At the low level, we proposed a novel distributed motion planning algorithm that addresses the vehicle deadlocks in local planning and then quickly generates a feasible new velocity for the non-holonomic vehicle in dense environments, guaranteeing that each vehicle efficiently visits its assigned target positions. Extensive simulation experiments in large-scale scenarios for non-holonomic vehicles and two real-world experiments demonstrate the effectiveness and advantages of our method in practical applications. The source code of our method can be available at https://github.com/wuuya1/LRGO.Note to Practitioners—The motivation for this article stems from the need to solve the multi-vehicle task assignment and motion planning (MVTAMP) problem for non-holonomic vehicles in dense environments. Many real-world applications exist, such as clearing threat targets, aid rescue, and package delivery. However, when vehicles need to continuously visit a series of assigned targets, motion planning for non-holonomic vehicles becomes more difficult because it is more likely to occur sharp turns between adjacent target path nodes. In this case, a better task allocation scheme can often lead to more efficient target visits and save all vehicles’ total traveling distance. To bridge this, we proposed a hierarchical method for solving the MVTAMP problem in large-scale complex scenarios. The numerous large-scale simulations and two real-world experiments show the effectiveness of the proposed method. Our future work will focus on the integrated task assignment and motion planning problem for non-holonomic vehicles in highly dynamic scenarios.

To build commercial robots, skid-steering mechanical design is of increased popularity due to its manufacturing simplicity and unique mechanism. However, these also cause significant challenges on software and algorithm design, especially for the pose estimation (i.e., determining the robot’s rotation and position) of skid-steering robots, since they change their orientation with an inevitable skid. To tackle this problem, we propose a probabilistic sliding-window estimator dedicated to skid-steering robots, using measurements from a monocular camera, the wheel encoders, and optionally an inertial measurement unit (IMU). Specifically, we explicitly model the kinematics of skid-steering robots by both track instantaneous centers of rotation (ICRs) and correction factors, which are capable of compensating for the complexity of track-to-terrain interaction, the imperfectness of mechanical design, terrain conditions and smoothness, etc. To prevent performance reduction in robots’ long-term missions, the time- and location- varying kinematic parameters are estimated online along with pose estimation states in a tightly-coupled manner. More importantly, we conduct in-depth observability analysis for different sensors and design configurations in this paper, which provides us with theoretical tools in making the correct choice when building real commercial robots. In our experiments, we validate the proposed method by both simulation tests and real-world experiments, which demonstrate that our method outperforms competing methods by wide margins. Note to Practitioners—This paper was motivated by the problem of long-term pose estimation of the commonly commercial-used skid-steering robots with only low-cost sensors. Skid-steering robots change their orientation with a skid, which poses a significant challenge for pose estimation when using the wheel encoders. We propose to online estimate the robot’s kinematics, which succeeds in compensating for the complexity of track-to-terrain interaction, due to the slippage, the imperfectness of mechanical design, terrain conditions and smoothness. It is critical to estimate the kinematics and poses jointly to prevent performance reduction in robots’ long-term missions. We further theoretically analyze whether the kinematics parameters can be estimated under different sensor configurations, and find out the special degrade motions that make the parameters unobservable.

The time series self-supervised contrastive learning framework has succeeded significantly in industrial fault detection scenarios. It typically consists of pretraining on abundant unlabeled data and fine-tuning on limited annotated data. However, the two-phase framework faces three challenges: Sampling bias, task-agnostic representation issue, and angular-centricity issue. These challenges hinder further development in industrial applications. This article introduces a debiased contrastive learning with supervision guidance (DCLSG) framework and applies it to industrial fault detection tasks. First, DCLSG employs channel augmentation to integrate temporal and frequency domain information. Pseudolabels based on momentum clustering operation are assigned to extracted representations, thereby mitigating the sampling bias raised by the selection of positive pairs. Second, the generated supervisory signal guides the pretraining phase, tackling the task-agnostic representation issue. Third, the angular-centricity issue is addressed using the proposed Gaussian distance metric measuring the radial distribution of representations. The experiments conducted on three industrial datasets (ISDB, CWRU, and practical datasets) validate the superior performance of the DCLSG compared to other fault detection methods.

Building reliable fault detection systems through deep neural networks is an appealing topic in industrial scenarios. In these contexts, the representations extracted by neural networks on available labeled time-series data can reflect system states. However, this endeavor remains challenging due to the necessity of labeled data. Self-supervised contrastive learning (SSCL) is one of the effective approaches to deal with this challenge, but existing SSCL-based models suffer from sampling bias and representation bias problems. This article introduces a debiased contrastive learning framework for time-series data and applies it to industrial fault detection tasks. This framework first develops the multigranularity augmented view generation method to generate augmented views at different granularities. It then introduces the momentum clustering contrastive learning strategy and the expert knowledge guidance mechanism to mitigate sampling bias and representation bias, respectively. Finally, the experiments on a public bearing fault detection dataset and a widely used valve stiction detection dataset show the effectiveness of the proposed feature learning framework.

Semantic scene completion (SSC) aims to predict the semantic occupancy of each voxel in the entire 3D scene from limited observations, which is an emerging and critical task for autonomous driving. Recently, many studies have turned to camera-based SSC solutions due to the richer visual cues and cost-effectiveness of cameras. However, existing methods usually rely on sophisticated and heavy 3D models to process the lifted 3D features directly, which are not discriminative enough for clear segmentation boundaries. In this paper, we adopt the dense-sparse-dense design and propose a one-stage camera-based SSC framework, termed SGN, to propagate semantics from the semantic-aware seed voxels to the whole scene based on spatial geometry cues. Firstly, to exploit depth-aware context and dynamically select sparse seed voxels, we redesign the sparse voxel proposal network to process points generated by depth prediction directly with the coarse-to-fine paradigm. Furthermore, by designing hybrid guidance (sparse semantic and geometry guidance) and effective voxel aggregation for spatial geometry cues, we enhance the feature separation between different categories and expedite the convergence of semantic propagation. Finally, we devise the multi-scale semantic propagation module for flexible receptive fields while reducing the computation resources. Extensive experimental results on the SemanticKITTI and SSCBench-KITTI-360 datasets demonstrate the superiority of our SGN over existing state-of-the-art methods. And even our lightweight version SGN-L achieves notable scores of 14.80% mIoU and 45.45% IoU on SeamnticKITTI validation with only 12.5 M parameters and 7.16 G training memory. Code is available at https://github.com/Jieqianyu/SGN.

Dense depth recovery is crucial in autonomous driving, serving as a foundational element for obstacle avoidance, 3D object detection, and local path planning. Adverse weather conditions, including haze, dust, rain, snow, and darkness, introduce significant challenges to accurate dense depth estimation, thereby posing substantial safety risks in autonomous driving. These challenges are particularly pronounced for traditional depth estimation methods that rely on short electromagnetic wave sensors, such as visible spectrum cameras and near-infrared LiDAR, due to their susceptibility to diffraction noise and occlusion in such environments. To fundamentally overcome this issue, we present a novel approach for robust metric depth estimation by fusing a millimeter-wave radar and a monocular infrared thermal camera, which are capable of penetrating atmospheric particles and unaffected by lighting conditions. Our proposed Radar-Infrared fusion method achieves highly accurate and finely detailed dense depth estimation through three stages, including monocular depth prediction with global scale alignment, quasi-dense radar augmentation by learning radar-pixels correspondences, and local scale refinement of dense depth using a scale map learner. Our method achieves exceptional visual quality and accurate metric estimation by addressing the challenges of ambiguity and misalignment that arise from directly fusing multi-modal long-wave features. We evaluate the performance of our approach on the NTU4DRadLM dataset and our self-collected challenging ZJU-Multispectrum dataset. Especially noteworthy is the unprecedented robustness demonstrated by our proposed method in smoky scenarios. Our code will be released athttps://github.com/MMOCKING/RIDERS.

Deep neural models have achieved remarkable performance on various supervised and unsupervised learning tasks, but it is a challenge to deploy these large-size networks on resource-limited devices. As a representative type of model compression and acceleration methods, knowledge distillation (KD) solves this problem by transferring knowledge from heavy teachers to lightweight students. However, most distillation methods focus on imitating the responses of teacher networks but ignore the information redundancy of student networks. In this article, we propose a novel distillation framework difference-based channel contrastive distillation (DCCD), which introduces channel contrastive knowledge and dynamic difference knowledge into student networks for redundancy reduction. At the feature level, we construct an efficient contrastive objective that broadens student networks' feature expression space and preserves richer information in the feature extraction stage. At the final output level, more detailed knowledge is extracted from teacher networks by making a difference between multiview augmented responses of the same instance. We enhance student networks to be more sensitive to minor dynamic changes. With the improvement of two aspects of DCCD, the student network gains contrastive and difference knowledge and reduces its overfitting and redundancy. Finally, we achieve surprising results that the student approaches and even outperforms the teacher in test accuracy on CIFAR-100. We reduce the top-1 error to 28.16% on ImageNet classification and 24.15% for cross-model transfer with ResNet-18. Empirical experiments and ablation studies on popular datasets show that our proposed method can achieve state-of-the-art accuracy compared with other distillation methods.

Recently value-based centralized training with decentralized execution (CTDE) multi-agent reinforcement learning (MARL) methods have achieved excellent performance in cooperative tasks. However, the most representative method among these methods, Q-network MIXing (QMIX), restricts the joint action Q values to be a monotonic mixing of each agent's utilities. Furthermore, current methods cannot generalize to unseen environments or different agent configurations, which is known as ad hoc team play situation. In this work, we propose a novel Q values decomposition that considers both the return of an agent acting on its own and cooperating with other observable agents to address the nonmonotonic problem. Based on the decomposition, we propose a greedy action searching method that can improve exploration and is not affected by changes in observable agents or changes in the order of agents' actions. In this way, our method can adapt to ad hoc team play situation. Furthermore, we utilize an auxiliary loss related to environmental cognition consistency and a modified prioritized experience replay (PER) buffer to assist training. Our extensive experimental results show that our method achieves significant performance improvements in both challenging monotonic and nonmonotonic domains, and can handle the ad hoc team play situation perfectly.

Spatial and temporal modeling is one of the most core aspects of few-shot action recognition. Most previous works mainly focus on long-term temporal relation modeling based on high-level spatial representations, without considering the crucial low-level spatial features and short-term temporal relations. Actually, the former feature could bring rich local semantic information, and the latter feature could represent motion characteristics of adjacent frames, respectively. In this paper, we propose SloshNet, a new framework that revisits the spatial and temporal modeling for few-shot action recognition in a finer manner. First, to exploit the low-level spatial features, we design a feature fusion architecture search module to automatically search for the best combination of the low-level and high-level spatial features. Next, inspired by the recent transformer, we introduce a long-term temporal modeling module to model the global temporal relations based on the extracted spatial appearance features. Meanwhile, we design another short-term temporal modeling module to encode the motion characteristics between adjacent frame representations. After that, the final predictions can be obtained by feeding the embedded rich spatial-temporal features to a common frame-level class prototype matcher. We extensively validate the proposed SloshNet on four few-shot action recognition datasets, including Something-Something V2, Kinetics, UCF101, and HMDB51. It achieves favorable results against state-of-the-art methods in all datasets.

Most of the existing blind image Super-Resolution (SR) methods assume that the blur kernels are space-invariant. However, the blur involved in real applications are usually space-variant due to object motion, out-of-focus, etc., resulting in severe performance drop of the advanced SR methods. To address this problem, we firstly introduce two new datasets with out-of-focus blur, i.e., NYUv2-BSR and Cityscapes-BSR, to support further researches of blind SR with space-variant blur. Based on the datasets, we design a novel Cross-MOdal fuSion network (CMOS) that estimate both blur and semantics simultaneously, which leads to improved SR results. It involves a feature Grouping Interactive Attention (GIA) module to make the two modalities in-teract more effectively and avoid inconsistency. GIA can also be used for the interaction of other features because of the universality of its structure. Qualitative and quantitative experiments compared with state-of-the-art methods on above datasets and real-world images demonstrate the superiority of our method, e.g., obtaining PSNR/SSIM by +1.91↑/+0.0048↑ on NYUv2-BSR than MANet11https://github.com/ByChelsea/CMOS.git.

For point cloud reconstruction-related tasks, the reconstruction losses to evaluate the shape differences between reconstructed results and the ground truths are typically used to train the task networks. Most existing works measure the training loss with point-to-point distance, which may introduce extra defects as predefined matching rules may deviate from the real shape differences. Although some learning-based works have been proposed to overcome the weaknesses of manually-defined rules, they still measure the shape differences in 3D Euclidean space, which may limit their ability to capture defects in reconstructed shapes. In this work, we propose a learning-based Contrastive Adver-sarial Loss (CALoss) to measure the point cloud reconstruction loss dynamically in a non-linear representation space by combining the contrastive constraint with the adversarial strategy. Specifically, we use the contrastive constraint to help CALoss learn a representation space with shape similarity, while we introduce the adversarial strategy to help CALoss mine differences between reconstructed results and ground truths. According to experiments on reconstruction-related tasks, CALoss can help task networks improve re-construction performances and learn more representative representations.

Recently, emotional talking face generation has received considerable attention. However, existing methods only adopt one-hot coding, image, or audio as emotion conditions, thus lacking flexible control in practical applications and failing to handle unseen emotion styles due to limited semantics. They either ignore the one-shot setting or the quality of generated faces. In this paper, we propose a more flexible and generalized framework. Specifically, we supplement the emotion style in text prompts and use an Aligned Multi-modal Emotion encoder to embed the text, image, and audio emotion modality into a unified space, which inherits rich semantic prior from CLIP. Consequently, effective multi-modal emotion space learning helps our method support arbitrary emotion modality during testing and could generalize to unseen emotion styles. Besides, an Emotion-aware Audio-to-3DMM Convertor is proposed to connect the emotion condition and the audio sequence to structural representation. A followed style-based High-fidelity Emotional Face generator is designed to generate arbitrary high-resolution realistic identities. Our texture generator hierarchically learns flow fields and animated faces in a residual manner. Extensive experiments demonstrate the flexibility and generalization of our method in emotion control and the effectiveness of high-quality face synthesis.

Structured pruning and quantization are promising approaches for reducing the inference time and memory footprint of neural networks. However, most existing methods require the original training dataset to fine-tune the model. This not only brings heavy resource consumption but also is not possible for applications with sensitive or proprietary data due to privacy and security concerns. Therefore, a few data-free methods are proposed to address this problem, but they perform data-free pruning and quantization separately, which does not explore the complementarity of pruning and quantization. In this paper, we propose a novel framework named Unified Data-Free Compression(UDFC), which performs pruning and quantization simultaneously without any data and fine-tuning process. Specifically, UDFC starts with the assumption that the partial information of a damaged(e.g., pruned or quantized) channel can be preserved by a linear combination of other channels, and then derives the reconstruction form from the assumption to restore the information loss due to compression. Finally, we formulate the reconstruction error between the original network and its compressed network, and theoretically deduce the closed-form solution. We evaluate the UDFC on the large-scale image classification task and obtain significant improvements over various network architectures and compression methods. For example, we achieve a 20.54% accuracy improvement on ImageNet dataset compared to SOTA method with 30% pruning ratio and 6-bit quantization on ResNet-34. Code will be available at here.

Recently, neural implicit surfaces have become popular for multi-view reconstruction. To facilitate practical applications like scene editing and manipulation, some works extend the framework with semantic masks input for the object-compositional reconstruction rather than the holistic perspective. Though achieving plausible disentanglement, the performance drops significantly when processing the indoor scenes where objects are usually partially observed. We propose RICO to address this by regularizing the unobservable regions for indoor compositional reconstruction. Our key idea is to first regularize the smoothness of the occluded background, which then in turn guides the foreground object reconstruction in unobservable regions based on the object-background relationship. Particularly, we regularize the geometry smoothness of occluded background patches. With the improved background surface, the signed distance function and the reversedly rendered depth of objects can be optimized to bound them within the background range. Extensive experiments show our method outperforms other methods on synthetic and real-world indoor scenes and prove the effectiveness of proposed regularizations. The code is available at https://github.com/kyleleey/RICO

Siamese network has been a de facto benchmark framework for 3D LiDAR object tracking with a shared-parametric encoder extracting features from template and search region, respectively. This paradigm relies heavily on an additional matching network to model the cross-correlation/similarity of the template and search region. In this paper, we forsake the conventional Siamese paradigm and propose a novel single-branch framework, SyncTrack, synchronizing the feature extracting and matching to avoid forwarding encoder twice for template and search region as well as introducing extra parameters of matching network. The synchronization mechanism is based on the dynamic affinity of the Transformer, and an in-depth analysis of the relevance is provided theoretically. Moreover, based on the synchronization, we introduce a novel Attentive PointsSampling strategy into the Transformer layers (APST), replacing the random/Farthest Points Sampling (FPS) method with sampling under the supervision of attentive relations between the template and search region. It implies connecting point-wise sampling with the feature learning, beneficial to aggregating more distinctive and geometric features for tracking with sparse points. Extensive experiments on two benchmark datasets (KITTI and NuScenes) show that SyncTrack achieves state-of-the-art performance in realtime tracking.

Image harmonization aims to solve the visual inconsistency problem in composited images by adaptively adjusting the foreground pixels with the background as references. Existing methods employ local color transformation or region matching between foreground and background, which neglects powerful proximity prior and independently distinguishes fore-/back-ground as a whole part for harmonization. As a result, they still show a limited performance across varied foreground objects and scenes. To address this issue, we propose a novel Global-aware Kernel Net-work (GKNet) to harmonize local regions with comprehensive consideration of long-distance background references. Specifically, GKNet includes two parts, i.e., harmony kernel prediction and harmony kernel modulation branches. The former includes a Long-distance Reference Extractor (LRE) to obtain long-distance context and Kernel Prediction Blocks (KPB) to predict multi-level harmony kernels by fusing global information with local features. To achieve this goal, a novel Selective Correlation Fusion (SCF) module is proposed to better select relevant long-distance background references for local harmonization. The latter employs the predicted kernels to harmonize foreground regions with local and global awareness. Abundant experiments demonstrate the superiority of our method for image harmonization over state-of-the-art methods, e.g., achieving 39.53dB PSNR that surpasses the best counterpart by +0.78dB ↑; decreasing fMSE/MSE by 11.5%↓/6.7%↓ compared with the SoTA method. Code will be available at here.

Class prototype construction and matching are core aspects of few-shot action recognition. Previous methods mainly focus on designing spatiotemporal relation modeling modules or complex temporal alignment algorithms. Despite the promising results, they ignored the value of class prototype construction and matching, leading to unsatisfactory performance in recognizing similar categories in every task. In this paper, we propose GgHM, a new framework with Graph-guided Hybrid Matching. Concretely, we learn task-oriented features by the guidance of a graph neural network during class prototype construction, optimizing the intra- and inter-class feature correlation explicitly. Next, we design a hybrid matching strategy, combining frame-level and tuple-level matching to classify videos with multivariate styles. We additionally propose a learnable dense temporal modeling module to enhance the video feature temporal representation to build a more solid foundation for the matching process. GgHM shows consistent improvements over other challenging baselines on several few-shot datasets, demonstrating the effectiveness of our method. The code will be publicly available at https://github.com/jiazheng-xing/GgHM.

Pairwise point cloud registration is a critical task for many applications, which heavily depends on finding correct correspondences from the two point clouds. However, the low overlap between input point clouds causes the registration to fail easily, leading to mistaken overlapping and mismatched correspondences, especially in scenes where non-overlapping regions contain similar structures. In this paper, we present a unified bird's-eye view (BEV) model for jointly learning of 3D local features and overlap estimation to fulfill pairwise registration and loop closure. Feature description is performed by a sparse UNet-like network based on BEV representation, and 3D keypoints are extracted by a detection head for 2D locations, and a regression head for heights. For overlap detection, a cross-attention module is applied for interacting contextual information of input point clouds, followed by a classification head to estimate the overlapping region. We evaluate our unified model extensively on the KITTI dataset and Apollo-SouthBay dataset. The experiments demonstrate that our method significantly outperforms existing methods on overlap estimation, especially in scenes with small overlaps. It also achieves top registration performance on both datasets in terms of translation and rotation errors.

Multi-sensor fusion-based localization technology has achieved high accuracy in autonomous systems. How to improve the robustness is the main challenge at present. The most commonly used LiDAR and camera are weather-sensitive, while the FMCW radar has strong adaptability but suffers from noise and ghost effects. In this paper, we propose a heterogeneous localization method of Radar on LiDAR Map (RoLM), which can eliminate the accumulated error of radar odometry in real-time to achieve higher localization accuracy without dependence on loop closures. We embed the two sensor modalities into a density map and calculate the spatial vector similarity with offset to seek the corresponding place index in the candidates and calculate the rotation and translation. We use the ICP to pursue perfect matching on the LiDAR submap based on the coarse alignment. Extensive experiments on Mulran Radar Dataset, Oxford Radar RobotCar Dataset, and our data verify the feasibility and effectiveness of our approach.

This paper investigates the problem of fixed-wing unmanned aerial vehicles (UAV s) motion planning with posture constraints and the problem of the more general symmetrical situations where UAVs have more than one optimal solution. In this paper, the posture constraints are formulated in the 3D Dubins method, and the symmetrical situations are overcome by a more collaborative strategy called the shunted strategy. The effectiveness of the proposed method has been validated by conducting extensive simulation experiments. Meanwhile, we compared the proposed method with the other state-of-the-art methods, and the comparison results show that the proposed method advances the previous works. Finally, the practicability of the proposed algorithm was analyzed by the statistic in computational cost. The source code of our method can be available at https://github.com/wuuya1/SCA.

The great potential of unsupervised monocular depth estimation has been demonstrated by many works due to low annotation cost and impressive accuracy comparable to supervised methods. To further improve the performance, recent works mainly focus on designing more complex network structures and exploiting extra supervised information, e.g., semantic segmentation. These methods optimize the models by exploiting the reconstructed relationship between the target and reference images in varying degrees. However, previous methods prove that this image reconstruction optimization is prone to get trapped in local minima. In this paper, our core idea is to guide the optimization with prior knowledge from pretrained Flow-Net. And we show that the bottleneck of unsupervised monocular depth estimation can be broken with our simple but effective framework named FG-Depth. In particular, we propose (i) a flow distillation loss to replace the typical photometric loss that limits the capacity of the model and (ii) a prior flow based mask to remove invalid pixels that bring the noise in training loss. Extensive experiments demonstrate the effectiveness of each component, and our approach achieves state-of-the-art results on both KITTI and NYU-Depth-v2 datasets.

Highly accurate and efficient map in unknown and complex environments is essential for robotics navigation. Traditionally, mobile robot platforms are often computationally constrained when using multiple sensors to process large amounts of input data. In previous works, some of them have been deployed to embedded platforms in real-time. However, how to balance accuracy and efficiency while reducing the computational resources and the memory footprint is still the bottleneck. Motivated by these challenges, we proposed a mapping framework called VDBblox to incrementally build Euclidean Signed Distance Fields (ESDFs) map from Truncated Signed Distance Fields (TSDFs) mapping. We use a novel weight function to update the non-projective TSDFs, thus improving the quality of the mesh reconstruction with higher accuracy than up-to-date methods. Meanwhile, the generated ESDFs map is maintained by the least recently used (LRU) cache to dynamically handle the obstacle changes with less runtime than state-of-the-art. We show VDBblox performance in terms of accuracy and efficiency by benchmark comparison on RGB-D and LiDAR public datasets. Moreover, we demonstrate that VDBblox can be integrated into a completed quadrotor system as a sub-module. Then we validate it through online obstacle avoidance and high-quality mesh reconstruction in real-world experiments. Finally, we release our method as open-source code to the community11Code - https://github.com/yinloonga/vdbblox.

This paper proposes a LiDAR-Inertial SLAM with efficiently extracted planes, which couples explicit planes in the odometry to improve accuracy and in the mapping for consistency. The proposed method consists of three parts: an efficient Point$\boldsymbol{\rightarrow\text{Line}\rightarrow \text{Plane}}$extraction algorithm, a LiDAR-Inertial-Plane tightly coupled odometry, and a global plane-aided mapping. Specifically, we leverage the ring field of the LiDAR point cloud to accelerate the region-growing-based plane extraction algorithm. Then we tightly coupled IMU pre-integration factors, LiDAR odometry factors, and explicit plane factors in the sliding window to obtain a more accurate initial pose for mapping. Finally, we maintain explicit planes in the global map, and enhance system consistency by optimizing the factor graph of optimized odometry factors and plane observation factors. Experimental results show that our plane extraction method is efficient, and the proposed plane-aided LiDAR-Inertial SLAM significantly improves the accuracy and consistency compared to the other state-of-the-art algorithms with only a small increase in time consumption.

Reliable LiDAR panoptic segmentation (LPS), including both semantic and instance segmentation, is vital for many robotic applications, such as autonomous driving. This work proposes a new LPS framework named PANet to eliminate the dependency on the offset branch and improve the performance on large objects, which are always over-segmented by clustering algorithms. Firstly, we propose a non-learning Sparse Instance Proposal (SIP) module with the “sampling-shifting-grouping” scheme to directly group thing points into instances from the raw point cloud efficiently. More specifically, balanced point sampling is introduced to generate sparse seed points with more uniform point distribution over the distance range. And a shift module, termed bubble shifting, is proposed to shrink the seed points to the clustered centers. Then we utilize the connected component label algorithm to generate instance proposals. Furthermore, an instance aggregation module is devised to integrate potentially fragmented instances, improving the performance of the SIP module on large objects. Extensive experiments show that PANet achieves state-of-the-art performance among published works on the SemanticKITII validation and nuScenes validation for the panoptic segmentation task. Code is available at https://github.com/Jieqianyu/PANet.git.

Semantic scene completion (SSC) jointly predicts the semantics and geometry of the entire 3D scene, which plays an essential role in 3D scene understanding for autonomous driving systems. SSC has achieved rapid progress with the help of semantic context in segmentation. However, how to effectively exploit the relationships between the semantic context in semantic segmentation and geometric structure in scene completion remains under exploration. In this paper, we propose to solve outdoor SSC from the perspective of representation separation and BEV fusion. Specifically, we present the network, named SSC-RS, which uses separate branches with deep supervision to explicitly disentangle the learning procedure of the semantic and geometric representations. And a BEV fusion network equipped with the proposed Adaptive Representation Fusion (ARF) module is presented to aggregate the multi-scale features effectively and efficiently. Due to the low computational burden and powerful representation ability, our model has good generality while running in real-time. Extensive experiments on SemanticKITTI demonstrate our SSC-RS achieves state-of-the-art performance. Code is available at https://github.com/Jieqianyu/SSC-RS.git.

This paper examines a pursuit-evasion game (PEG) involving multiple pursuers and evaders. The decentralized pursuers aim to collaborate to capture the faster evaders while avoiding collisions. The policies of all agents are learning-based and are subjected to kinematic constraints that are specific to unicycles. To address the challenge of high dimensionality encountered in large-scale scenarios, we propose a state processing method named Mix-Attention, which is based on Self-Attention. This method effectively mitigates the curse of dimensionality. The simulation results provided in this study demonstrate that the combination of Mix-Attention and Independent Proximal Policy Optimization (IPPO) surpasses alternative approaches when solving the multi-pursuer multi-evader PEG, particularly as the number of entities increases. Moreover, the trained policies showcase their ability to adapt to scenarios involving varying numbers of agents and obstacles without requiring retraining. This adaptability showcases their transferability and robustness. Finally, our proposed approach has been validated through physical experiments conducted with six robots.

An event camera is a novel vision sensor that can capture per-pixel brightness changes and output a stream of asynchronous “events”. It has advantages over conventional cameras in those scenes with high-speed motions and challenging lighting conditions because of the high temporal resolution, high dynamic range, low bandwidth, low power consumption, and no motion blur. Therefore, several supervised monocular depth estimation from events is proposed to address scenes difficult for conventional cameras. However, depth annotation is costly and time-consuming. In this paper, to lower the annotation cost, we propose a self-supervised event-based monocular depth estimation framework named EMoDepth. EMoDepth constrains the training process using the cross-modal consistency from intensity frames that are aligned with events in the pixel coordinate. Moreover, in inference, only events are used for monocular depth prediction. Additionally, we design a multi-scale skip-connection architecture to effectively fuse features for depth estimation while maintaining high inference speed. Experiments on MVSEC and DSEC datasets demonstrate that our contributions are effective and that the accuracy can outperform existing supervised event-based and unsupervised frame-based methods.

This paper focuses on LiDAR Panoptic Segmentation (LPS), which has attracted more attention recently due to its broad application prospect for autonomous driving and robotics. The mainstream LPS approaches either adopt a top-down strategy relying on 3D object detectors to discover instances or utilize time-consuming heuristic clustering algorithms to group instances in a bottom-up manner. Inspired by the center representation and kernel-based segmentation, we propose a new detection-free and clustering-free framework called CenterLPS, with the center-based instance encoding and decoding paradigm. Specifically, we propose a sparse center proposal network to generate the sparse 3D instance centers, as well as center feature embedding, which can well encode characteristics of instances. Then a center-aware transformer is applied to collect the context between different center feature embedding and around centers. Moreover, we generate the kernel weights based on the enhanced center feature embedding and initialize dynamic convolutions to decode the final instance masks. Finally, a mask fusion module is devised to unify the semantic and instance predictions and improve the panoptic quality. Extensive experiments on SemanticKITTI and nuScenes demonstrate the effectiveness of our proposed center-based framework CenterLPS.

Real-world time series are characterized by intrinsic non-stationarity that poses a principal challenge for deep forecasting models. While previous models suffer from complicated series variations induced by changing temporal distribution, we tackle non-stationary time series with modern Koopman theory that fundamentally considers the underlying time-variant dynamics. Inspired by Koopman theory of portraying complex dynamical systems, we disentangle time-variant and time-invariant components from intricate non-stationary series by Fourier Filter and design Koopman Predictor to advance respective dynamics forward. Technically, we propose Koopa as a novel Koopman forecaster composed of stackable blocks that learn hierarchical dynamics. Koopa seeks measurement functions for Koopman embedding and utilizes Koopman operators as linear portraits of implicit transition. To cope with time-variant dynamics that exhibits strong locality, Koopa calculates context-aware operators in the temporal neighborhood and is able to utilize incoming ground truth to scale up forecast horizon. Besides, by integrating Koopman Predictors into deep residual structure, we ravel out the binding reconstruction loss in previous Koopman forecasters and achieve end-to-end forecasting objective optimization. Compared with the state-of-the-art model, Koopa achieves competitive performance while saving 77.3% training time and 76.0% memory.

Abstract As a special case of common object removal, image person removal is playing an increasingly important role in social media and criminal investigation domains. Due to the integrity of person area and the complexity of human posture, person removal has its own dilemmas. In this paper, a novel idea is proposed to tackle these problems from the perspective of data synthesis. Concerning the lack of a dedicated dataset for image person removal, two dataset production methods are proposed to automatically generate images, masks and ground truths, respectively. Then, a learning framework similar to local image degradation is proposed so that the masks can be used to guide the feature extraction process and more texture information can be gathered for final prediction. A coarse‐to‐fine training strategy is further applied to refine the details. The data synthesis and learning framework combine well with each other. Experimental results verify the effectiveness of the method quantitatively and qualitatively, and the trained network proves to have good generalization ability either on real or synthetic images.

Recent methods for action recognition always apply 3D Convolutional Neural Networks (CNNs) to extract spatiotemporal features and introduce optical flows to present motion features. Although achieving state-of-the-art performance, they are expensive in both time and space. In this paper, we propose to represent both two kinds of features in a unified 2D CNN without any 3D convolution or optical flows calculation. In particular, we first design a channel-wise spatiotemporal module to present the spatiotemporal features and a channel-wise motion module to encode feature-level motion features efficiently. Besides, we provide a distinctive illustration of the two modules from the frequency domain by interpreting them as advanced and learnable versions of frequency components. Second, we combine these two modules and an identity mapping path into one united block that can easily replace the original residual block in the ResNet architecture, forming a simple yet effective network dubbed STM network by introducing very limited extra computation cost and parameters. Third, we propose a novel Twins Training framework for action recognition by incorporating a correlation loss to optimize the inter-class and intra-class correlation and a siamese structure to fully stretch the training data. We extensively validate the proposed STM on both temporal-related datasets (i.e., Something-Something v1 & v2) and scene-related datasets (i.e., Kinetics-400, UCF-101, and HMDB-51). It achieves favorable results against state-of-the-art methods in all the datasets.

Density-based and classification-based methods have ruled unsupervised anomaly detection in recent years, while reconstruction-based methods are rarely mentioned for the poor reconstruction ability and low performance. However, the latter requires no costly extra training samples for the unsupervised training that is more practical, so this paper focuses on improving reconstruction-based method and proposes a novel O mni-frequency C hannel-selection R econstruction (OCR-GAN) network to handle sensory anomaly detection task in a perspective of frequency. Concretely, we propose a Frequency Decoupling (FD) module to decouple the input image into different frequency components and model the reconstruction process as a combination of parallel omni-frequency image restorations, as we observe a significant difference in the frequency distribution of normal and abnormal images. Given the correlation among multiple frequencies, we further propose a Channel Selection (CS) module that performs frequency interaction among different encoders by adaptively selecting different channels. Abundant experiments demonstrate the effectiveness and superiority of our approach over different kinds of methods, e.g., achieving a new state-of-the-art 98.3 detection AUC on the MVTec AD dataset without extra training data that markedly surpasses the reconstruction-based baseline by +38.1 ↑ and the current SOTA method by +0.3 ↑ . The source code is available in the additional materials.

In this article, we present a fast real-time tangled memory network that segments the objects effectively and efficiently for semi-supervised video object segmentation (VOS). We propose a tangled reference encoder and a memory bank organization mechanism based on a state estimator to fully utilize the mask features and alleviate memory overhead and computational burden brought by the unlimited memory bank used in many memory-based methods. First, the tangled memory network exploits the mask features that uncover abundant object information like edges and contours but are not fully explored in existing methods. Specifically, a tangled two-stream reference encoder is designed to extract and fuse the features from both RGB frames and the predicted masks. Second, to indicate the quality of the predicted mask and feedback the online prediction state for organizing the memory bank, we devise a target state estimator to learn the IoU score between the predicted mask and ground truth. Moreover, to accelerate the forward process and avoid memory overflow, we use a memory bank of fixed size to store historical features by designing a new efficient memory bank organization mechanism based on the mask state score provided by the state estimator. We conduct comprehensive experiments on the public benchmarks DAVIS and YouTube-VOS, demonstrating that our method obtains competitive results while running at high speed (66 FPS on the DAVIS16-val set).

Point cloud completion is an interesting and challenging task in 3D vision, which aims to recover complete shapes from sparse and incomplete point clouds. Existing completion networks often require a vast number of parameters and substantial computational costs to achieve a high performance level, which may limit their practical application. In this work, we propose a novel Adaptive efficient Recurrent Forward Network (ARFNet), which is composed of three parts: Recurrent Feature Extraction (RFE), Forward Dense Completion (FDC) and Raw Shape Protection (RSP). In an RFE, multiple short global features are extracted from incomplete point clouds, while a dense quantity of completed results are generated in a coarse-to-fine pipeline in the FDC. Finally, we propose the Adamerge module to preserve the details from the original models by merging the generated results with the original incomplete point clouds in the RSP. In addition, we introduce the Sampling Chamfer Distance to better capture the shapes of the models and the balanced expansion constraint to restrict the expansion distances from coarse to fine. According to the experiments on ShapeNet and KITTI, our network can achieve state-of-the-art completion performances on dense point clouds with fewer parameters, smaller model sizes, lower memory costs and a faster convergence.

In the practical application of restoring low-resolution gray-scale images, we generally need to run three separate processes of image colorization, super-resolution, and dows-sampling operation for the target device. However, this pipeline is redundant and inefficient for the independent processes, and some inner features could have been shared. Therefore, we present an efficient paradigm to perform Simultaneously Image Colorization and Super-resolution (SCS) and propose an end-to-end SCSNet to achieve this goal. The proposed method consists of two parts: colorization branch for learning color information that employs the proposed plug-and-play Pyramid Valve Cross Attention (PVCAttn) module to aggregate feature maps between source and reference images; and super-resolution branch for integrating color and texture information to predict target images, which uses the designed Continuous Pixel Mapping (CPM) module to predict high-resolution images at continuous magnification. Furthermore, our SCSNet supports both automatic and referential modes that is more flexible for practical application. Abundant experiments demonstrate the superiority of our method for generating authentic images over state-of-the-art methods, e.g., averagely decreasing FID by 1.8 and 5.1 compared with current best scores for automatic and referential modes, respectively, while owning fewer parameters (more than x2) and faster running speed (more than x3).

This paper presents a novel Region-Aware Face Swapping (RAFSwap) network to achieve identity-consistent harmonious high-resolution face generation in a local-global manner: 1) Local Facial Region-Aware (FRA) branch augments local identity-relevant features by introducing the Transformer to effectively model misaligned crossscale semantic interaction. 2) Global Source Feature-Adaptive (SFA) branch further complements global identity-relevant cues for generating identity-consistent swapped faces. Besides, we propose a Face Mask Predictor (FMP) module incorporated with StyleGAN2 to predict identity-relevant soft facial masks in an unsupervised manner that is more practical for generating harmonious high-resolution faces. Abundant experiments qualitatively and quantitatively demonstrate the superiority of our method for generating more identity-consistent high-resolution swapped faces over SOTA methods, e.g., obtaining 96.70 ID retrieval that outperforms SOTA MegaFS by$5.87\uparrow$.

This paper proposes an online large-scale dense mapping method for UAVs with a height of 150–250 meters. We first fuse the GPS with the visual odometry to estimate the scaled poses and sparse points. In order to use the depth of sparse points for depth map, we propose Sparse Confidence Cascade View-Aggregation MVSNet (SCCVA-MVSNet), which projects the depth-converged points in the sliding window on keyframes to obtain a sparse depth map. To weigh the confidence of the depth of each sparse point, we construct sparse confidence by the photometric error. The images of all keyframes, coarse depth, and confidence as the input of CVA-MVSNet to extract features and construct 3D cost volumes with adaptive view aggregation to balance the different stereo baselines between the keyframes. Our proposed network utilizes sparse features point information, the output of the network better maintains the consistency of the scale. Our experiments show that MVSNet using sparse feature point information outperforms image-only MVSNet, and our online reconstruction results are comparable to offline reconstruction methods. To benefit the research community, we open our code at https://github.com/hjxwhy/LODM.git

Transformers have shown great power in time series forecasting due to their global-range modeling ability. However, their performance can degenerate terribly on non-stationary real-world data in which the joint distribution changes over time. Previous studies primarily adopt stationarization to attenuate the non-stationarity of original series for better predictability. But the stationarized series deprived of inherent non-stationarity can be less instructive for real-world bursty events forecasting. This problem, termed over-stationarization in this paper, leads Transformers to generate indistinguishable temporal attentions for different series and impedes the predictive capability of deep models. To tackle the dilemma between series predictability and model capability, we propose Non-stationary Transformers as a generic framework with two interdependent modules: Series Stationarization and De-stationary Attention. Concretely, Series Stationarization unifies the statistics of each input and converts the output with restored statistics for better predictability. To address the over-stationarization problem, De-stationary Attention is devised to recover the intrinsic non-stationary information into temporal dependencies by approximating distinguishable attentions learned from raw series. Our Non-stationary Transformers framework consistently boosts mainstream Transformers by a large margin, which reduces MSE by 49.43% on Transformer, 47.34% on Informer, and 46.89% on Reformer, making them the state-of-the-art in time series forecasting. Code is available at this repository: https://github.com/thuml/Nonstationary_Transformers.

Abstract Dataset synthesis via virtual engines like Unity is attracting much more attention in recent years due to its low cost at obtaining ground‐truth labels. For this kind of work, virtual environments are constructed within the engine to mimic the real‐world, either with great manual efforts or learning‐based methods. The latter shows superiority over the former when the target real‐world scenes are changeable, from which the attributes of environments can be automatically adjusted based on the distribution difference between the synthetic and real‐world datasets. However, the non‐differentiability of whole pipeline hinders the efficiency of attribute optimization. To this end, this paper proposes to simulate synthetic datasets from a fine‐grained perspective, such that the system can be trained at an end‐to‐end manner. Specifically, it is converted into an image‐level data synthesis problem, and designs a constraint using the content loss between two images. As the rendering process of virtual engine is mathematically unknown, which blocks the back propagation of the gradients, a generative model is trained to approximate the engine. As a result, the whole framework becomes fully differentiable and the attributes can be optimized efficiently by gradient descent. Experimental result shows the efficiency of our method in obtaining useful synthetic training datasets. Besides, it is found that the image‐level method enables to learn the potential distribution of real‐world data, which is hard to be achieved by existing methods. As far as we know, it is the first attempt to finish this task with a differentiable process.

Model hubs with many pre-trained models (PTMs) have become a cornerstone of deep learning. Although built at a high cost, they remain under-exploited---practitioners usually pick one PTM from the provided model hub by popularity and then fine-tune the PTM to solve the target task. This naïve but common practice poses two obstacles to full exploitation of pre-trained model hubs: first, the PTM selection by popularity has no optimality guarantee, and second, only one PTM is used while the remaining PTMs are ignored. An alternative might be to consider all possible combinations of PTMs and extensively fine-tune each combination, but this would not only be prohibitive computationally but may also lead to statistical over-fitting. In this paper, we propose a new paradigm for exploiting model hubs that is intermediate between these extremes. The paradigm is characterized by two aspects: (1) We use an evidence maximization procedure to estimate the maximum value of label evidence given features extracted by pre-trained models. This procedure can rank all the PTMs in a model hub for various types of PTMs and tasks before fine-tuning. (2) The best ranked PTM can either be fine-tuned and deployed if we have no preference for the model's architecture or the target PTM can be tuned by the top $K$ ranked PTMs via a Bayesian procedure that we propose. This procedure, which we refer to as B-Tuning, not only improves upon specialized methods designed for tuning homogeneous PTMs, but also applies to the challenging problem of tuning heterogeneous PTMs where it yields a new level of benchmark performance.

Audio-guided face reenactment aims to generate authentic target faces that have matched facial expression of the input audio, and many learning-based methods have successfully achieved this. However, most methods can only reenact a particular person once trained or suffer from the low-quality generation of the target images. Also, nearly none of the current reenactment works consider the model size and running speed that are important for practical use. To solve the above challenges, we propose an efficientAudio-guidedMulti-face reenactment model namedAMNet, which can reenact target faces among multiple persons with corresponding source faces and drive signals as inputs. Concretely, we design aGeometric Controller(GC) module to inject the drive signals so that the model can be optimized in an end-to-end manner and generate more authentic images. Also, we adopt a lightweight network for our face reenactor so that the model can run in real-time on both CPU and GPU devices. Abundant experiments prove our approach’s superiority over existing methods,e.g., averagely decreasing FID by 0.12$\downarrow$and increasing SSIM by 0.031$\uparrow$than APB2Face, while owning fewer parameters ($\times 4 \downarrow$) and faster CPU speed ($\times 4 \uparrow$).

Video object detection (VOD) focuses on detecting objects for each frame in a video, which is a challenging task due to appearance deterioration in certain video frames. Recent works usually distill crucial information from multiple support frames to improve the reference features, but they only perform at frame level or proposal level that cannot integrate spatial-temporal features sufficiently. To deal with this challenge, we treat VOD as a spatial-temporal hierarchical features interacting process and introduce a Multi-level Spatial-Temporal (MST) feature aggregation framework to fully exploit frame-level, proposal-level, and instance-level information in a unified framework. Specifically, MST first measures context similarity in pixel space to enhance all frame-level features rather than only update reference features. The proposal-level feature aggregation then models object relation to augment reference object proposals. Furthermore, to filter out irrelevant information from other classes and backgrounds, we introduce an instance ID constraint to boost instance-level features by leveraging support object proposal features that belong to the same object. Besides, we propose a Deformable Feature Alignment (DAlign) module before MST to achieve a more accurate pixel-level spatial alignment for better feature aggregation. Extensive experiments are conducted on ImageNet VID and UAVDT datasets that demonstrate the superiority of our method over state-of-the-art (SOTA) methods. Our method achieves 83.3% and 62.1% with ResNet-101 on two datasets, outperforming SOTA MEGA by 0.4% and 2.7%.

This paper focuses on the mask utilization of video object segmentation (VOS). The mask here mains the reference masks in the memory bank, i.e., several chosen high-quality predicted masks, which are usually used with the reference frames together. The reference masks depict the edge and contour features of the target object and indicate the boundary of the target against the background, while the reference frames contain the raw RGB information of the whole image. It is obvious that the reference masks could play a significant role in the VOS, but this is not well explored yet. To tackle this, we propose to investigate the mask advantages of both the encoder and the matcher. For the encoder, we provide a unified codebase to integrate and compare eight different mask-fused encoders. Half of them are inherited or summarized from existing methods, and the other half are devised by ourselves. We find the best configuration from our design and give valuable observations from the comparison. Then, we propose a new mask-enhanced matcher to reduce the background distraction and enhance the locality of the matching process. Combining the mask-fused encoder, mask-enhanced matcher and a standard decoder, we formulate a new architecture named MaskVOS, which sufficiently exploits the mask benefits for VOS. Qualitative and quantitative results demonstrate the effectiveness of our method. We hope our exploration could raise the attention of mask utilization in VOS.

Accurate estimation of short-term traffic flow, which can help to assist travelers make better route choices, is a significant research field of intelligent transportation system. In order to extract complex spatiotemporal features from a small amount of available traffic data, in this paper we propose a novel Deep Graph Gaussian Processes (DGGPs) for short-term traffic flow prediction. First, in order to accurately describe the relationship between vertices in time series, this paper proposes an attention kernel. Based on this, the Aggregation Gaussian Process uses attention kernel as the covariance function, which overcomes the problem that the existing Gaussian processes and the deep Gaussian processes cannot effectively obtain dynamic spatial features. Second, DGGPs are constructed by the Aggregation Gaussian Process (AGP), the Temporal Convolutional Gaussian Process (TCGP) and the Gaussian process with linear kernel, to solve the existing short-term traffic flow forecasting models cannot obtain complex spatiotemporal features from a small amount of available data. We verify that the attention kernel helps to the proposed model convergence on the three data sets. At the same time, the proposed DGGP can obtain spatiotemporal features from the situation with less available spatial information or temporal information, accurately predict short-term traffic flow, and quantify temporal uncertainty.

Small object detection remains an unsolved challenge because it is hard to extract the information of small objects with only a few pixels. While scale-level corresponding detection in feature pyramid network alleviates this problem, we find feature coupling of various scales still impairs the performance of small objects. In this paper, we propose an extended feature pyramid network (EFPN) with an extra high-resolution pyramid level specialized for small object detection. Specifically, we design a novel module, named feature texture transfer (FTT), which is used to super-resolve features and extract credible regional details simultaneously. Moreover, we introduce a cross resolution distillation mechanism to transfer the ability of perceiving details across the scales of the network, where a foreground-background-balanced loss function is designed to alleviate area imbalance of foreground and background. In our experiments, the proposed EFPN is efficient on both computation and memory, and yields state-of-the-art results on small traffic-sign dataset Tsinghua-Tencent 100 K and small category of general object detection dataset MS COCO.

Since the development of 3D applications, the point cloud, as a spatial description easily acquired by sensors, has been widely used in multiple areas such as SLAM and 3D reconstruction. Point Cloud Compression (PCC) has also attracted more attention as a primary step before point cloud transferring and saving, where the geometry compression is an important component of PCC to compress the points geometrical structures. However, existing non-learning-based geometry compression methods are often limited by manually pre-defined compression rules. Though learning-based compression methods can significantly improve the algorithm performances by learning compression rules from data, they still have some defects. Voxel-based compression networks introduce precision errors due to the voxelized operations, while point-based methods may have relatively weak robustness and are mainly designed for sparse point clouds. In this work, we propose a novel learning-based point cloud compression framework named 3D Point Cloud Geometry Quantiation Compression Network (3QNet), which overcomes the robustness limitation of existing point-based methods and can handle dense points. By learning a codebook including common structural features from simple and sparse shapes, 3QNet can efficiently deal with multiple kinds of point clouds. According to experiments on object models, indoor scenes, and outdoor scans, 3QNet can achieve better compression performances than many representative methods.

Accurate and reliable sensor calibration is essential to fuse LiDAR and inertial measurements, which are usually available in robotic applications. In this article, we propose a novel LiDAR-IMU calibration method within the continuous-time batch-optimization framework, where the intrinsics of both sensors and the spatial-temporal extrinsics between sensors are calibrated without using calibration infrastructure, such as fiducial tags. Compared to discrete-time approaches, the continuous-time formulation has natural advantages for fusing high-rate measurements from LiDAR and IMU sensors. To improve efficiency and address degenerate motions, the following two observability-aware modules are leveraged: first, The information-theoretic data selection policy selectsonlythe most informative segments for calibration during data collection, which significantly improves the calibration efficiency by processing only the selected informative segments. Second, the observability-aware state update mechanism in nonlinear least-squares optimization updatesonlythe identifiable directions in the state space with truncated singular value decomposition, which enables accurate calibration results even under degenerate cases where informative data segments are not available. The proposed LiDAR-IMU calibration approach has been validated extensively in both simulated and real-world experiments with different robot platforms, demonstrating its high accuracy and repeatability in commonly-seen human-made environments.

Extracting the specific category of the players, such as the malignant Bot, from the huge log data of the massive multiplayer online role playing games, denoted as MMORPGs, is an important basic task in game security and personal recommendation. In this article, we propose a parallel semisupervised framework to categorize specific game players with a few label-known target samples, which are denoted as bait players. Our approach first presents a feature representation model based on the players’ level granularity, which can acquire aligned feature representations in the lower dimensional space from the players’ original action sequences. Then, we propose a semisupervised clustering method, extended from the bisecting$k$-means model, to extract the specified players with the help of those bait players. Due to massive amounts of game log data, the computation complexity is an extreme challenge to implement our feature representation and semisupervised extraction approaches. We also propose a hierarchical parallelism framework, which allows the data to be computed horizontally and vertically simultaneously and enables varied parallel combinations for the steps of our semisupervised categorization approach. The comparable experiments on real-world MMORPGs’ log data, containing more than 465 Gbytes and million players, are carried out to demonstrate the effectiveness and efficiency of our proposed approach compared with the state-of-the-art methods.

Depth completion aims to recover a dense depth map from a sparse depth map with the corresponding color image as input. Recent approaches mainly formulate the depth completion as a one-stage end-to-end learning task, which outputs dense depth maps directly. However, the feature extraction and supervision in one-stage frameworks are insufficient, limiting the performance of these approaches. To address this problem, we propose a novel end-to-end residual learning framework, which formulates the depth completion as a two-stage learning task, i.e., a sparse-to-coarse stage and a coarse-to-fine stage. First, a coarse dense depth map is obtained by a simple CNN framework. Then, a refined depth map is further obtained using a residual learning strategy in the coarse-to-fine stage with coarse depth map and color image as input. Specially, in the coarse-to-fine stage, a channel shuffle extraction operation is utilized to extract more representative features from color image and coarse depth map, and an energy based fusion operation is exploited to effectively fuse these features obtained by channel shuffle operation, thus leading to more accurate and refined depth maps. We achieve SoTA performance in RMSE on KITTI benchmark. Extensive experiments on other datasets future demonstrate the superiority of our approach over current state-of-the-art depth completion approaches.

Self-supervised learning shows great potential in monocular depth estimation, using image sequences as the only source of supervision. Although people try to use the high-resolution image for depth estimation, the accuracy of prediction has not been significantly improved. In this work, we find the core reason comes from the inaccurate depth estimation in large gradient regions, making the bilinear interpolation error gradually disappear as the resolution increases. To obtain more accurate depth estimation in large gradient regions, it is necessary to obtain high-resolution features with spatial and semantic information. Therefore, we present an improved DepthNet, HR-Depth, with two effective strategies: (1) re-design the skip-connection in DepthNet to get better high-resolution features and (2) propose feature fusion Squeeze-and-Excitation(fSE) module to fuse feature more efficiently. Using Resnet-18 as the encoder, HR-Depth surpasses all previous state-of-the-art(SoTA) methods with the least parameters at both high and low resolution. Moreover, previous SoTA methods are based on fairly complex and deep networks with a mass of parameters which limits their real applications. Thus we also construct a lightweight network which uses MobileNetV3 as encoder. Experiments show that the lightweight network can perform on par with many large models like Monodepth2 at high-resolution with only20%parameters. All codes and models will be available at https://github.com/shawLyu/HR-Depth.

In this paper we tackle the problem of pose guided person image generation, which aims to transfer a person image from the source pose to a novel target pose while maintaining the source appearance. Given the inefficiency of standard CNNs in handling large spatial transformation, we propose a structure-aware flow based method for high-quality person image generation. Specifically, instead of learning the complex overall pose changes of human body, we decompose the human body into different semantic parts (e.g., head, torso, and legs) and apply different networks to predict the flow fields for these parts separately. Moreover, we carefully design the network modules to effectively capture the local and global semantic correlations of features within and among the human parts respectively. Extensive experimental results show that our method can generate high-quality results under large pose discrepancy and outperforms state-of-the-art methods in both qualitative and quantitative comparisons.

The paper proposes a novel generative adversarial network for one-shot face reenactment, which can animate a single face image to a different pose-and-expression (provided by a driving image) while keeping its original appearance. The core of our network is a novel mechanism called appearance adaptive normalization, which can effectively integrate the appearance information from the input image into our face generator by modulating the feature maps of the generator using the learned adaptive parameters. Furthermore, we specially design a local net to reenact the local facial components (i.e., eyes, nose and mouth) first, which is a much easier task for the network to learn and can in turn provide explicit anchors to guide our face generator to learn the global appearance and pose-and-expression. Extensive quantitative and qualitative experiments demonstrate the significant efficacy of our model compared with prior one-shot methods.

Point cloud completion is an interesting and challenging task in 3D vision, aiming to recover complete shapes from sparse and incomplete point clouds. Existing learning-based methods often require vast computation cost to achieve excellent performance, which limits their practical applications. In this paper, we propose a novel Recurrent Forward Network (RFNet), which is composed of three modules: Recurrent Feature Extraction (RFE), Forward Dense Completion (FDC) and Raw Shape Protection (RSP). The RFE extracts multiple global features from the incomplete point clouds for different recurrent levels, and the FDC generates point clouds in a coarse-to-fine pipeline. The RSP introduces details from the original incomplete models to refine the completion results. Besides, we propose a Sampling Chamfer Distance to better capture the shapes of models and a new Balanced Expansion Constraint to restrict the expansion distances from coarse to fine. According to the experiments on ShapeNet and KITTI, our network can achieve the state-of-the-art with lower memory cost and faster convergence.

Remarkable results have been achieved by DCNN based self-supervised depth estimation approaches. However, most of these approaches can only handle either day-time or night-time images, while their performance degrades for all-day images due to large domain shift and the variation of illumination between day and night images. To relieve these limitations, we propose a domain-separated network for self-supervised depth estimation of all-day images. Specifically, to relieve the negative influence of disturbing terms (illumination, etc.), we partition the information of day and night image pairs into two complementary sub-spaces: private and invariant domains, where the former contains the unique information (illumination, etc.) of day and night images and the latter contains essential shared information (texture, etc.). Meanwhile, to guarantee that the day and night images contain the same information, the domain-separated network takes the day-time images and corresponding night-time images (generated by GAN) as input, and the private and invariant feature extractors are learned by orthogonality and similarity loss, where the domain gap can be alleviated, thus better depth maps can be expected. Meanwhile, the reconstruction and photometric losses are utilized to estimate complementary information and depth maps effectively. Experimental results demonstrate that our approach achieves state-of-the-art depth estimation results for all-day images on the challenging Oxford RobotCar dataset, proving the superiority of our proposed approach. Code and data split are available at https://github.com/LINA-lln/ADDS-DepthNet.

This paper studies task adaptive pre-trained model selection, an underexplored problem of assessing pre-trained models for the target task and select best ones from the model zoo \emph{without fine-tuning}. A few pilot works addressed the problem in transferring supervised pre-trained models to classification tasks, but they cannot handle emerging unsupervised pre-trained models or regression tasks. In pursuit of a practical assessment method, we propose to estimate the maximum value of label evidence given features extracted by pre-trained models. Unlike the maximum likelihood, the maximum evidence is \emph{immune to over-fitting}, while its expensive computation can be dramatically reduced by our carefully designed algorithm. The Logarithm of Maximum Evidence (LogME) can be used to assess pre-trained models for transfer learning: a pre-trained model with a high LogME value is likely to have good transfer performance. LogME is \emph{fast, accurate, and general}, characterizing itself as the first practical method for assessing pre-trained models. Compared with brute-force fine-tuning, LogME brings at most $3000\times$ speedup in wall-clock time and requires only $1%$ memory footprint. It outperforms prior methods by a large margin in their setting and is applicable to new settings. It is general enough for diverse pre-trained models (supervised pre-trained and unsupervised pre-trained), downstream tasks (classification and regression), and modalities (vision and language). Code is available at this repository: \href{https://github.com/thuml/LogME}{https://github.com/thuml/LogME}.

In this paper, we present PocoNet: Point cloud Online COmpression NETwork to address the task of SLAM-oriented compression. The aim of this task is to select a compact subset of points with high priority to maintain localization accuracy. The key insight is that points with high priority have similar geometric features in SLAM scenarios. Hence, we tackle this task as point cloud segmentation to capture complex geometric information. We calculate observation counts by matching between maps and point clouds and divide them into different priority levels. Trained by labels annotated with such observation counts, the proposed network could evaluate the point-wise priority. Experiments are conducted by integrating our compression module into an existing SLAM system to evaluate compression ratios and localization performances. Experimental results on two different datasets verify the feasibility and generalization of our approach.

LiDAR-based SLAM system is admittedly more accurate and stable than others, while its loop closure detection is still an open issue. With the development of 3D semantic segmentation for point cloud, semantic information can be obtained conveniently and steadily, essential for high-level intelligence and conductive to SLAM. In this paper, we present a novel semantic-aided LiDAR SLAM with loop closure based on LOAM, named SA-LOAM, which leverages semantics in odometry as well as loop closure detection. Specifically, we propose a semantic-assisted ICP, including semantically matching, downsampling and plane constraint, and integrates a semantic graph-based place recognition method in our loop closure detection module. Benefitting from semantics, we can improve the localization accuracy, detect loop closures effectively, and construct a global consistent semantic map even in large-scale scenes. Extensive experiments on KITTI and Ford Campus dataset show that our system significantly improves baseline performance, has generalization ability to unseen data and achieves competitive results compared with state-of-the-art methods.

In this work, we present a lightweight, tightly-coupled deep depth network and visual-inertial odometry (VIO) system, which can provide accurate state estimates and dense depth maps of the immediate surroundings. Leveraging the proposed lightweight Conditional Variational Autoencoder (CVAE) for depth inference and encoding, we provide the network with previously marginalized sparse features from VIO to increase the accuracy of initial depth prediction and generalization capability. The compact representation of dense depth, termed depth code, can be updated jointly with navigation states in a sliding window estimator in order to provide the dense local scene geometry. We additionally propose a novel method to obtain the CVAE’s Jacobian which is shown to be more than an order of magnitude faster than previous works, and we additionally leverage First-Estimate Jacobian (FEJ) to avoid recalculation. As opposed to previous works that rely on completely dense residuals, we propose to only provide sparse measurements to update the depth code and show through careful experimentation that our choice of sparse measurements and FEJs can still significantly improve the estimated depth maps. Our full system also exhibits state-of-the-art pose estimation accuracy, and we show that it can run in real-time with single-thread execution while utilizing GPU acceleration only for the network and code Jacobian.

Place recognition gives a SLAM system the ability to correct cumulative errors. Unlike images that contain rich texture features, point clouds are almost pure geometric information which makes place recognition based on point clouds challenging. Existing works usually encode low-level features such as coordinate, normal, reflection intensity, etc., as local or global descriptors to represent scenes. Besides, they often ignore the translation between point clouds when matching descriptors. Different from most existing methods, we explore the use of high-level features, namely semantics, to improve the descriptor’s representation ability. Also, when matching descriptors, we try to correct the translation between point clouds to improve accuracy. Concretely, we propose a novel global descriptor, Semantic Scan Context, which explores semantic information to represent scenes more effectively. We also present a two-step global semantic ICP to obtain the 3D pose (x, y, yaw) used to align the point cloud to improve matching performance. Our experiments on the KITTI dataset show that our approach outperforms the state-of-the- art methods with a large margin. Our code is available at: https://github.com/lilin-hitcrt/SSC.

Multi-agent path finding in formation has many potential real-world applications like mobile warehouse robotics. However, previous multi-agent path finding (MAPF) methods hardly take formation into consideration. Further-more, they are usually centralized planners and require the whole state of the environment. Other decentralized partially observable approaches to MAPF are reinforcement learning (RL) methods. However, these RL methods encounter difficulties when learning path finding and formation problems at the same time. In this paper, we propose a novel decentralized partially observable RL algorithm that uses a hierarchical structure to decompose the multi-objective task into unrelated ones. It also calculates a theoretical weight that makes each tasks reward has equal influence on the final RL value function. Additionally, we introduce a communication method that helps agents cooperate with each other. Experiments in simulation show that our method outperforms other end-to-end RL methods and our method can naturally scale to large world sizes where centralized planner struggles. We also deploy and validate our method in a real-world scenario.

In this paper, we propose a highly accurate continuous-time trajectory estimation framework dedicated to SLAM (Simultaneous Localization and Mapping) applications, which enables fuse high-frequency and asynchronous sensor data effectively. We apply the proposed framework in a 3D LiDAR-inertial system for evaluations. The proposed method adopts a non-rigid registration method for continuous-time trajectory estimation and simultaneously removing the motion distortion in LiDAR scans. Additionally, we propose a two-state continuous-time trajectory correction method to efficiently and efficiently tackle the computationally-intractable global optimization problem when loop closure happens. We examine the accuracy of the proposed approach on several publicly available datasets and the data we collected. The experimental results indicate that the proposed method outperforms the discrete-time methods regarding accuracy especially when aggressive motion occurs. Furthermore, we open source our code at https://github.com/APRIL-ZJU/clins to benefit research community.

Outdoor scene completion is a challenging issue in 3D scene understanding, which plays an important role in intelligent robotics and autonomous driving. Due to the sparsity of LiDAR acquisition, it is far more complex for 3D scene completion and semantic segmentation. Since semantic features can provide constraints and semantic priors for completion tasks, the relationship between them is worth exploring. Therefore, we propose an end-to-end semantic segmentation-assisted scene completion network, including a 2D completion branch and a 3D semantic segmentation branch. Specifically, the network takes a raw point cloud as input, and merges the features from the segmentation branch into the completion branch hierarchically to provide semantic information. By adopting BEV representation and 3D sparse convolution, we can benefit from the lower operand while maintaining effective expression. Besides, the decoder of the segmentation branch is used as an auxiliary, which can be discarded in the inference stage to save computational consumption. Extensive experiments demonstrate that our method achieves competitive performance on SemanticKITTI dataset with low latency. Code and models will be released at https://github.com/jokester-zzz/SSA-SC.

An efficient 3D scene perception algorithm is a vital component for autonomous driving and robotics systems. In this paper, we focus on semantic scene completion, which is a task of jointly estimating the volumetric occupancy and semantic labels of objects. Since the real-world data is sparse and occluded, this is an extremely challenging task. We propose a novel framework, named Up-to-Down network (UDNet), to achieve the large-scale semantic scene completion with an encoder-decoder architecture for voxel grids. The novel up-to-down block can effectively aggregate multi-scale context information to improve labeling coherence, and the atrous spatial pyramid pooling module is leveraged to expand the receptive field while preserving detailed geometric information. Besides, the proposed multi-scale fusion mechanism efficiently aggregates global background information and improves the semantic completion accuracy. Moreover, to further satisfy the needs of different tasks, our UDNet can accomplish the multi-resolution semantic completion, achieving faster but coarser completion. Detailed experiments in the semantic scene completion benchmark of SemanticKITTI illustrate that our proposed framework surpasses the state-of-the-art methods with remarkable margins and a real-time inference speed by using only voxel grids as input.

Currently, there have been many kinds of pointbased 3D trackers, while voxel-based methods are still underexplored. In this paper, we first propose a voxel-based tracker, named PointSiamRCNN, improving tracking performance by embedding target information into the search region. Our framework is composed of two parts for achieving proposal generation and proposal refinement, which fully releases the potential of the two-stage object tracking. Specifically, it takes advantage of efficient feature learning of the voxel-based Siamese network and high-quality proposal generation of the Siamese region proposal network head. In the search region, the groundtruth annotations are utilized to realize semantic segmentation, which leads to more discriminative feature learning with pointwise supervisions. Furthermore, we propose the Self and Cross Attention Module for embedding target information into the search region. Finally, the multi-scale RoI pooling module is proposed to obtain compact representations from target-aware features for proposal refinement. Exhaustive experiments on the KITTI tracking dataset demonstrate that our framework reaches the competitive performance with the state-of-the-art 3D tracking methods and achieves the state-of-the-art in terms of BEV tracking.

Inspired by biological evolution, we explain the rationality of Vision Transformer by analogy with the proven practical Evolutionary Algorithm (EA) and derive that both of them have consistent mathematical representation. Analogous to the dynamic local population in EA, we improve the existing transformer structure and propose a more efficient EAT model, and design task-related heads to deal with different tasks more flexibly. Moreover, we introduce the spatial-filling curve into the current vision transformer to sequence image data into a uniform sequential format. Thus we can design a unified EAT framework to address multi-modal tasks, separating the network architecture from the data format adaptation. Our approach achieves state-of-the-art results on the ImageNet classification task compared with recent vision transformer works while having smaller parameters and greater throughput. We further conduct multi-modal tasks to demonstrate the superiority of the unified EAT, \eg, Text-Based Image Retrieval, and our approach improves the rank-1 by +3.7 points over the baseline on the CSS dataset.

Filter pruning for a pre-trained convolutional neural network is most normally performed through human-made constraints or criteria such as norms, ranks, etc. Typically, the pruning pipeline comprises two-stage: first learn a sparse structure from the original model, then optimize the weights in the new prune model. One disadvantage of using human-made criteria to prune filters is that the design and selection of threshold criteria depend on complicated prior knowledge. Besides, the pruning process is less robust due to the impact of directly regularizing on filters. To address the problems mentioned, we propose an effective one-stage pruning framework: introducing a trainable collaborative layer to jointly prune and learn neural networks in one go. In our framework, we first add a binary collaborative layer for each original filter. Then, a new type of gradient estimator - asymptotic gradient estimator is first introduced to pass the gradient in the binary collaborative layer. Finally, we simultaneously learn the sparse structure and optimize the weights from the original model in the training process. Our evaluation results on typical benchmarks, CIFAR and ImageNet, demonstrate very promising results against other state-of-the-art filter pruning methods.

Due to the high representational efficiency, superpixel largely reduces the number of image primitives for subsequent processing. However, superpixel is scarcely utilized in recent methods since its irregular shape is intractable for standard convolutional layer. In this paper, we propose an end-to-end trainable superpixel convolutional network, named SPNet, to learn high-level representation on image superpixel primitives. We start by treating irregular superpixel lattices as a 2D point cloud, where the low-level features inside one superpixel are aggregated to one feature vector. We replace the standard convolutional layer with the PointConv layer to handle the irregular and unordered point cloud. Besides, we propose grid based downsampling strategies to output uniform 2D sampling result. The resulting network largely utilizes the efficiency of superpixel and provides a novel view for image recognition task. Experiments on image recognition task show promising results compared with prominent image classification methods. The visualization of class activation mapping shows great accuracy at object localization and boundary segmentation.

This paper addresses the guided depth completion task in which the goal is to predict a dense depth map given a guidance RGB image and sparse depth measurements. Recent advances on this problem nurture hopes that one day we can acquire accurate and dense depth at a very low cost. A major challenge of guided depth completion is to effectively make use of extremely sparse measurements, e.g., measurements covering less than 1% of the image pixels. In this paper, we propose a fully differentiable model that avoids convolving on sparse tensors by jointly learning depth interpolation and refinement. More specifically, we propose a differentiable kernel regression layer that interpolates the sparse depth measurements via learned kernels. We further refine the interpolated depth map using a residual depth refinement layer which leads to improved performance compared to learning absolute depth prediction using a vanilla network. We provide experimental evidence that our differentiable kernel regression layer not only enables end-to-end training from very sparse measurements using standard convolutional network architectures, but also leads to better depth interpolation results compared to existing heuristically motivated methods. We demonstrate that our method outperforms many state-of-the-art guided depth completion techniques on both NYUv2 and KITTI. We further show the generalization ability of our method with respect to the density and spatial statistics of the sparse depth measurements.

The quantized neural network (QNN) is an efficient approach for network compression and can be widely used in the implementation of field-programmable gate arrays (FPGAs). This article proposes a novel learning framework for n -bit QNNs, whose weights are constrained to the power of two. To solve the gradient vanishing problem, we propose a reconstructed gradient function for QNNs in the back-propagation algorithm that can directly get the real gradient rather than estimating an approximate gradient of the expected loss. We also propose a novel QNN structure named n -BQ-NN, which uses shift operation to replace the multiply operation and is more suitable for the inference on FPGAs. Furthermore, we also design a shift vector processing element (SVPE) array to replace all 16-bit multiplications with SHIFT operations in convolution operation on FPGAs. We also carry out comparable experiments to evaluate our framework. The experimental results show that the quantized models of ResNet, DenseNet, and AlexNet through our learning framework can achieve almost the same accuracies with the original full-precision models. Moreover, when using our learning framework to train our n -BQ-NN from scratch, it can achieve state-of-the-art results compared with typical low-precision QNNs. Experiments on Xilinx ZCU102 platform show that our n -BQ-NN with our SVPE can execute 2.9 times faster than that with the vector processing element (VPE) in inference. As the SHIFT operation in our SVPE array will not consume digital signal processing (DSP) resources on FPGAs, the experiments have shown that the use of SVPE array also reduces average energy consumption to 68.7% of the VPE array with 16 bit.

In this article, we focus on pose estimation dedicated to nonholonomic ground robots with low-cost sensors, by probabilistically fusing measurements from wheel odometers and an exteroceptive sensor. For ground robots, wheel odometers are widely used in pose estimation tasks, especially in applications in planar scenes. However, since wheel odometer only provides two-dimensional (2D) motion measurements, it is extremely challenging to use that for accurate full 6-D pose (3-D position and 3-D orientation) estimation. Traditional methods for 6-D pose estimation with wheel odometers either approximate motion profiles at the cost of accuracy reduction, or rely on other sensors, e.g., inertial measurement unit, to provide complementary measurements. By contrast, we propose a novel motion-manifold-based method for pose estimation of ground robots, which enables to utilize wheel odometers for high-precision 6-D pose estimation. Specifically, the proposed method, first, formulates the motion manifold of ground robots by a parametric representation, second, performs manifold-based 6-D integration with wheel odometer measurements only, and third, reparameterizes manifold representation periodically for error reduction. To demonstrate the effectiveness and applicability of the proposed algorithmic module, we integrate that into a sliding-window pose estimator by using measurements from wheel odometers and a monocular camera. Extensive simulated and real-world experiments are conducted for evaluation, and the proposed algorithm is shown to outperform competing the state-of-the-art algorithms by a significant margin in pose estimation accuracy, especially when deployed in complex, large-scale real-world environments.

In large-scale multi-agent systems, the large number of agents and complex game relationship cause great difficulty for policy learning. Therefore, simplifying the learning process is an important research issue. In many multi-agent systems, the interactions between agents often happen locally, which means that agents neither need to coordinate with all other agents nor need to coordinate with others all the time. Traditional methods attempt to use pre-defined rules to capture the interaction relationship between agents. However, the methods cannot be directly used in a large-scale environment due to the difficulty of transforming the complex interactions between agents into rules. In this paper, we model the relationship between agents by a complete graph and propose a novel game abstraction mechanism based on two-stage attention network (G2ANet), which can indicate whether there is an interaction between two agents and the importance of the interaction. We integrate this detection mechanism into graph neural network-based multi-agent reinforcement learning for conducting game abstraction and propose two novel learning algorithms GA-Comm and GA-AC. We conduct experiments in Traffic Junction and Predator-Prey. The results indicate that the proposed methods can simplify the learning process and meanwhile get better asymptotic performance compared with state-of-the-art algorithms.

A lot of efforts have been devoted to investigating how agents can learn effectively and achieve coordination in multiagent systems. However, it is still challenging in large-scale multiagent settings due to the complex dynamics between the environment and agents and the explosion of state-action space. In this paper, we design a novel Dynamic Multiagent Curriculum Learning (DyMA-CL) to solve large-scale problems by starting from learning on a multiagent scenario with a small size and progressively increasing the number of agents. We propose three transfer mechanisms across curricula to accelerate the learning process. Moreover, due to the fact that the state dimension varies across curricula, and existing network structures cannot be applied in such a transfer setting since their network input sizes are fixed. Therefore, we design a novel network structure called Dynamic Agent-number Network (DyAN) to handle the dynamic size of the network input. Experimental results show that DyMA-CL using DyAN greatly improves the performance of large-scale multiagent learning compared with state-of-the-art deep reinforcement learning approaches. We also investigate the influence of three transfer mechanisms across curricula through extensive simulations.

Recent works have shown how realistic talking face images can be obtained under the supervision of geometry guidance, e.g., facial landmark or boundary. To alleviate the demand for manual annotations, in this paper, we propose a novel self-supervised hybrid model (DAE-GAN) that learns how to reenact face naturally given large amounts of unlabeled videos. Our approach combines two deforming autoencoders with the latest advances in the conditional generation. On the one hand, we adopt the deforming autoencoder to disentangle identity and pose representations. A strong prior in talking face videos is that each frame can be encoded as two parts: one for video-specific identity and the other for various poses. Inspired by that, we utilize a multi-frame deforming autoencoder to learn a pose-invariant embedded face for each video. Meanwhile, a multi-scale deforming autoencoder is proposed to extract pose-related information for each frame. On the other hand, the conditional generator allows for enhancing fine details and overall reality. It leverages the disentangled features to generate photo-realistic and pose-alike face images. We evaluate our model on VoxCeleb1 and RaFD dataset. Experiment results demonstrate the superior quality of reenacted images and the flexibility of transferring facial movements between identities.

Head pose estimation from RGB images without depth information is a challenging task due to the loss of spatial information as well as large head pose variations in the wild. The performance of existing landmark-free methods remains unsatisfactory as the quality of estimated pose is inferior. In this paper, we propose a novel three-branch network architecture, termed as Feature Decoupling Network (FDN), a more powerful architecture for landmark-free head pose estimation from a single RGB image. In FDN, we first propose a feature decoupling (FD) module to explicitly learn the discriminative features for each pose angle by adaptively recalibrating its channel-wise responses. Besides, we introduce a cross-category center (CCC) loss to constrain the distribution of the latent variable subspaces and thus we can obtain more compact and distinct subspaces. Extensive experiments on both in-the-wild and controlled environment datasets demonstrate that the proposed method outperforms other state-of-the-art methods based on a single RGB image and behaves on par with approaches based on multimodal input resources.

Unsupervised learning of optical flow, which leverages the supervision from view synthesis, has emerged as a promising alternative to supervised methods. However, the objective of unsupervised learning is likely to be unreliable in challenging scenes. In this work, we present a framework to use more reliable supervision from transformations. It simply twists the general unsupervised learning pipeline by running another forward pass with transformed data from augmentation, along with using transformed predictions of original data as the self-supervision signal. Besides, we further introduce a lightweight network with multiple frames by a highly-shared flow decoder. Our method consistently gets a leap of performance on several benchmarks with the best accuracy among deep unsupervised methods. Also, our method achieves competitive results to recent fully supervised methods while with much fewer parameters.

This paper presents a novel multi-identity face reenactment framework, named FReeNet, to transfer facial expressions from an arbitrary source face to a target face with a shared model. The proposed FReeNet consists of two parts: Unified Landmark Converter (ULC) and Geometry-aware Generator (GAG). The ULC adopts an encode-decoder architecture to efficiently convert expression in a latent landmark space, which significantly narrows the gap of the face contour between source and target identities. The GAG leverages the converted landmark to reenact the photorealistic image with a reference image of the target person. Moreover, a new triplet perceptual loss is proposed to force the GAG module to learn appearance and geometry information simultaneously, which also enriches facial details of the reenacted images. Further experiments demonstrate the superiority of our approach for generating photorealistic and expression-alike faces, as well as the flexibility for transferring facial expressions between identities.

Audio-guided face reenactment aims at generating photorealistic faces using audio information while maintaining the same facial movement as when speaking to a real person. However, existing methods can not generate vivid face images or only reenact low-resolution faces, which limits the application value. To solve those problems, we propose a novel deep neural network named APB2Face, which consists of GeometryPredictor and FaceReenactor modules. GeometryPredictor uses extra head pose and blink state signals as well as audio to predict the latent landmark geometry information, while FaceReenactor inputs the face landmark image to reenact the photorealistic face. A new dataset AnnV I collected from YouTube is presented to support the approach, and experimental results indicate the superiority of our method than state-of-the-arts, whether in authenticity or controllability.

Due to the difficulty in generating the effective descriptors which are robust to occlusion and viewpoint changes, place recognition for 3D point cloud remains an open issue. Unlike most of the existing methods that focus on extracting local, global, and statistical features of raw point clouds, our method aims at the semantic level that can be superior in terms of robustness to environmental changes. Inspired by the perspective of humans, who recognize scenes through identifying semantic objects and capturing their relations, this paper presents a novel semantic graph based approach for place recognition. First, we propose a novel semantic graph representation for the point cloud scenes by reserving the semantic and topological information of the raw point cloud. Thus, place recognition is modeled as a graph matching problem. Then we design a fast and effective graph similarity network to compute the similarity. Exhaustive evaluations on the KITTI dataset show that our approach is robust to the occlusion as well as viewpoint changes and outperforms the state-of-the-art methods with a large margin. Our code is available at: https://github.com/kxhit/SG_PR.

Sensor calibration is the fundamental block for a multi-sensor fusion system. This paper presents an accurate and repeatable LiDAR-IMU calibration method (termed LI-Calib), to calibrate the 6-DOF extrinsic transformation between the 3D LiDAR and the Inertial Measurement Unit (IMU). Regarding the high data capture rate for LiDAR and IMU sensors, LI-Calib adopts a continuous-time trajectory formulation based on B-Spline, which is more suitable for fusing high-rate or asynchronous measurements than discrete-time based approaches. Additionally, LI-Calib decomposes the space into cells and identifies the planar segments for data association, which renders the calibration problem well-constrained in usual scenarios without any artificial targets. We validate the proposed calibration approach on both simulated and real-world experiments. The results demonstrate the high accuracy and good repeatability of the proposed method in common human-made scenarios. To benefit the research community, we open-source our code at https://github.com/APRIL-ZJU/lidar_IMU_calib.

This paper presents F-Siamese Tracker, a novel approach for single object tracking prominently characterized by more robustly integrating 2D and 3D information to reduce redundant search space. A main challenge in 3D single object tracking is how to reduce search space for generating appropriate 3D candidates. Instead of solely relying on 3D proposals, firstly, our method leverages the Siamese network applied on RGB images to produce 2D region proposals which are then extruded into 3D viewing frustums. Besides, we perform an on-line accuracy validation on the 3D frustum to generate refined point cloud searching space, which can be embedded directly into the existing 3D tracking backbone. For efficiency, our approach gains better performance with fewer candidates by reducing search space. In addition, benefited from introducing the online accuracy validation, for occasional cases with strong occlusions or very sparse points, our approach can still achieve high precision, even when the 2D Siamese tracker loses the target. This approach allows us to set a new state-of-the-art in 3D single object tracking by a significant margin on a sparse outdoor dataset (KITTI tracking). Moreover, experiments on 2D single object tracking show that our framework boosts 2D tracking performance as well.

Multi-sensor fusion of multi-modal measurements from commodity inertial, visual and LiDAR sensors to provide robust and accurate 6DOF pose estimation holds great potential in robotics and beyond. In this paper, building upon our prior work (i.e., LIC-Fusion), we develop a sliding-window filter based LiDAR-Inertial-Camera odometry with online spatiotemporal calibration (i.e., LIC-Fusion 2.0), which introduces a novel sliding-window plane-feature tracking for efficiently processing 3D LiDAR point clouds. In particular, after motion compensation for LiDAR points by leveraging IMU data, low-curvature planar points are extracted and tracked across the sliding window. A novel outlier rejection criteria is proposed in the plane-feature tracking for high quality data association. Only the tracked planar points belonging to the same plane will be used for plane initialization, which makes the plane extraction efficient and robust. Moreover, we perform the observability analysis for the IMU-LiDAR subsystem under consideration and report the degenerate cases for spatiotemporal calibration using plane features. While the estimation consistency and identified degenerate motions are validated in Monte-Carlo simulations, different real-world experiments are also conducted to show that the proposed LIC-Fusion 2.0 outperforms its predecessor and other state-of-the-art methods.

Self-localization is an essential technology for autonomous vehicles. Building robust odometry in a GPS-denied environment is still challenging, especially when LiDAR and camera are uninformative. In this paper, We propose a learning-based approach to cure the drift of gyroscope for vehicle localization. For consumer-level MEMS gyroscope (stability ~10° /h), our GyroNet can estimate the error of each measurement. For high-precision Fiber optics Gyroscope (stability ~0.05° /h), we build a FoGNet which can obtain its drift by observing data in a long time window. We perform comparative experiments on publicly available datasets. The results demonstrate that our GyroNet can get higher precision angular velocity than traditional digital filters and static initialization methods. In the vehicle localization, the FoGNet can effectively correct the small drift of the Fiber optics Gyroscope (FoG) and can achieve better results than the state-of-the-art method.

Character customization system is an important component in Role-Playing Games (RPGs), where players are allowed to edit the facial appearance of their in-game characters with their own preferences rather than using default templates. This paper proposes a method for automatically creating in-game characters of players according to an input face photo. We formulate the above "artistic creation" process under a facial similarity measurement and parameter searching paradigm by solving an optimization problem over a large set of physically meaningful facial parameters. To effectively minimize the distance between the created face and the real one, two loss functions, i.e. a "discriminative loss" and a "facial content loss", are specifically designed. As the rendering process of a game engine is not differentiable, a generative network is further introduced as an "imitator" to imitate the physical behavior of the game engine so that the proposed method can be implemented under a neural style transfer framework and the parameters can be optimized by gradient descent. Experimental results demonstrate that our method achieves a high degree of generation similarity between the input face photo and the created in-game character in terms of both global appearance and local details. Our method has been deployed in a new game last year and has now been used by players over 1 million times.

This paper presents a tightly-coupled aided inertial navigation system (INS) with point and plane features, a general sensor fusion framework applicable to any visual and depth sensor (e.g., RGBD, LiDAR) configuration, in which the camera is used for point feature tracking and depth sensor for plane extraction. The proposed system exploits geometrical structures (planes) of the environments and adopts the closest point (CP) for plane parameterization. Moreover, we distinguish planar point features from non-planar point features in order to enforce point-on-plane constraints which are used in our state estimator, thus further exploiting structural information from the environment. We also introduce a simple but effective plane feature initialization algorithm for feature-based simultaneous localization and mapping (SLAM). In addition, we perform online spatial calibration between the IMU and the depth sensor as it is difficult to obtain this critical calibration parameter in high precision. Both Monte-Carlo simulations and real-world experiments are performed to validate the proposed approach.

Many real-world problems, such as robot control and soccer game, are naturally modeled as sparse-interaction multi-agent systems. Reutilizing single-agent knowledge in multi-agent systems with sparse interactions can greatly accelerate the multi-agent learning process. Previous works rely on bisimulation metric to define Markov decision process (MDP) similarity for controlling knowledge transfer. However, bisimulation metric is costly to compute and is not suitable for high-dimensional state space problems. In this work, we propose more scalable transfer learning methods based on a novel MDP similarity concept. We start by defining the MDP similarity based on the N-step return (NSR) values of an MDP. Then, we propose two knowledge transfer methods based on deep neural networks called direct value function transfer and NSR-based value function transfer. We conduct experiments in image-based grid world, multi-agent particle environment (MPE) and Ms. Pac-Man game. The results indicate that the proposed methods can significantly accelerate multi-agent reinforcement learning and meanwhile get better asymptotic performance.

Scene flow estimation in the dynamic scene remains a challenging task. Computing scene flow by a combination of 2D optical flow and depth has shown to be considerably faster with acceptable performance. In this work, we present a unified framework for joint unsupervised learning of stereo depth and optical flow with explicit local rigidity to estimate scene flow. We estimate camera motion directly by a Perspective-n-Point method from the optical flow and depth predictions, with RANSAC outlier rejection scheme. In order to disambiguate the object motion and the camera motion in the scene, we distinguish the rigid region by the re-project error and the photometric similarity. By joint learning with the local rigidity, both depth and optical networks can be refined. This framework boosts all four tasks: depth, optical flow, camera motion estimation, and object motion segmentation. Through the evaluation on the KITTI benchmark, we show that the proposed framework achieves state-of-the-art results amongst unsupervised methods. Our models and code are available at https://github.com/lliuz/unrigidflow.

In this paper, we propose PASS3D to achieve point-wise semantic segmentation for 3D point cloud. Our framework combines the efficiency of traditional geometric methods with robustness of deep learning methods, consisting of two stages: At stage -1, our accelerated cluster proposal algorithm will generate refined cluster proposals by segmenting point clouds without ground, capable of generating less redundant proposals with higher recall in an extremely short time; stage -2 we will amplify and further process these proposals by a neural network to estimate semantic label for each point and meanwhile propose a novel data augmentation method to enhance the network's recognition capability for all categories especially for non-rigid objects. Evaluated on KITTI raw dataset, PASS3D stands out against the state-of-the-art on some results, making itself competent to 3D perception in autonomous driving system. Our source code will be open-sourced. A video demonstration is available at https://www.youtube.com/watch?v=cukEqDuP_Qw.

In this paper, we present a robust and efficient estimation approach with multi-camera, odometer and gyroscope. Robust initialization, tightly-coupled optimization estimator and multi-camera loop-closure detection are utilized in the proposed approach. In initialization, the measurements of odometer and gyroscope are used to compute scale, and then estimate the bias of sensors. In estimator, the pre-integration of odometer and gyroscope is derived and combined with the measurements of multi-camera to estimate the motion in a tightly-coupled optimization framework. In loop-closure detection, a connection between different cameras of the vehicle can be built, which significantly improve the success rate of loop-closure detection. The proposed algorithm is validated in multiple real-world datasets collected in different places, time, weather and illumination. Experimental results show that the proposed approach can estimate the motion of vehicles robustly and efficiently.

We propose a hybrid pyramid based approach to fuse the direct and indirect methods in visual SLAM, to allow robust localization under various situations including large-baseline motion, low-texture environment, and various illumination changes. In our approach, we first calculate coarse inter-frame pose estimation by matching the feature points. Subsequently, we use both direct image alignment and a multiscale pyramid method, for refining the previous estimation to attain better precision. Furthermore, we perform online photometric calibration along with pose estimation, to reduce un-modelled errors. To evaluate our approach, we conducted various real-world experiments on both public datasets and self-collected ones, by implementing a full SLAM system with the proposed methods. The results show that our system improves both localization accuracy and robustness by a wide margin.

This paper presents a tightly-coupled multi-sensor fusion algorithm termed LiDAR-inertial-camera fusion (LIC-Fusion), which efficiently fuses IMU measurements, sparse visual features, and extracted LiDAR points. In particular, the proposed LIC-Fusion performs online spatial and temporal sensor calibration between all three asynchronous sensors, in order to compensate for possible calibration variations. The key contribution is the optimal (up to linearization errors) multi-modal sensor fusion of detected and tracked sparse edge/surf feature points from LiDAR scans within an efficient MSCKF-based framework, alongside sparse visual feature observations and IMU readings. We perform extensive experiments in both indoor and outdoor environments, showing that the proposed LIC-Fusion outperforms the state-of-the-art visual-inertial odometry (VIO) and LiDAR odometry methods in terms of estimation accuracy and robustness to aggressive motions.

In face-to-face conversations, head motions play a crucial role in encoding information, and humans are very skilled at decoding multiple messages from interlocutors' head motions. It is of great importance to endow embodied conversational agents (ECAs) with the capability of conveying communicative intention through head movements. Our work is aimed at automatically synthesizing head motions for an ECA speaking Chinese. We propose to take only transcripts as input to compute head movements, based on a statistical framework. Subjective experiments are conducted to validate the proposed statistical framework. The results show that the generated head animation is able to improve human perception in terms of naturalness and demonstrate that the head animation is synchronized with the input of synthetic speech.

3D point cloud presentation has been widely used in computer vision, automatic driving, augmented reality, smart cities and virtual reality. 3D point cloud compression method with higher compression ratio and tiny loss is the key to improve data transportation efficiency. In this paper, we propose a new 3D point cloud geometry compression method based on deep learning, also an auto-encoder performing better than other networks in detail reconstruction. It can reach much higher compression ratio than the state-of-art while keeping tolerable loss. It also supports parallel compressing multiple models by GPU, which can improve processing efficiency greatly. The compression process is composed of two parts. Firstly, Raw data is compressed into codeword by extracting feature of raw model with encoder. Then, the codeword is further compressed with sparse coding. Decompression process is implemented in reverse order. Codeword is recovered and fed into decoder to reconstruct point cloud. Detail reconstruction ability is improved by a hierarchical structure in our decoder. Latter outputs are grown from former fuzzier outputs. In this way, details are added to former output by latter layers step by step to make a more precise prediction. We compare our method with PCL compression and Draco compression on ShapeNet40 part dataset. Our method may be the first deep learning-based point cloud compression algorithm. The experiments demonstrate it is superior to former common compression algorithms with large compression ratio, which can also reserve original shapes with tiny loss.

We propose an active learning-based grasp method for accurate industrial manipulation that combines the high accuracy of geometrically driven grasp methods and the generalization ability of data-driven grasp methods. Our grasp sequence consists of pregrasp stage and grasp stage which integrates the active perception and manipulation. In pregrasp stage, the manipulator actively moves and perceives the object. At each step, given the perception image, a motion is chosen so that the manipulator can adjust to a proper pose to grasp the object. We train a convolutional neural network to estimate the motion and combine the network with a closed-loop control so that the end effector can move to the pregrasp state. In grasp stage, the manipulator executes a fixed motion to complete the grasp task. The fixed motion can be acquired from the demonstration with nonexpert conveniently. Our proposed method does not require the prior knowledge of camera intrinsic parameters, hand-eye transformation, or manually designed feature of objects. Instead, the training data sets containing prior knowledge are collected through interactive perception. The method can be easily transferred to new tasks with a few human interventions and is able to complete high accuracy grasp task with a certain robustness to partial observation condition. In our circuit board grasping tests, we could achieve a grasp accuracy of 0.8 mm and 0.6°.

Programming by demonstration (PBD) transforms the robot programming from the code level to automated interface between robot and human, promoting the flexibility of robotized automation. In this paper, we focus on programming the industrial robot for assembly tasks by parsing the human demonstration into a series of assembly skills and compiling the skill to the robot executables. To achieve this goal, an identification system using multimodal information to recognize the assembly skill, called MASD, is proposed including: 1) an initial learning stage using a hierarchical model to recognize the action by considering the features from action-object effect, gesture, and trajectory and 2) a retrospective thinking stage using a segmentation method to cut the continuous demonstrations into multiple assembly skills optimally. Using MASD, the demonstration of assembly tasks can be explained with high accuracy in real time, driving a hypothesis that a PBD system on the top of MASD can be extended to more realistic assembly tasks beyond pure positional moving and picking. In experiments, the skill identification module is used to recognize the five kinds of assembly skills in demonstrations of both single and multiple assembly skills, and outperforms the comparative action identification methods. Besides integrated with the MASD, the PBD system can generate the program based on the demonstration and successfully enable an ABB industrial robotic arm simulator to assemble a flashlight and a switch, verifying the initial hypothesis. Note to Practitioners-In the conventional robotized automation, the key role of the robot mainly owes to its capacity for repeating a wide variety of tasks with high speed and accuracy in long term, with a cost of days to months of programming for deployment. On the other hand, the new trend of customization brings the new characteristics: production in short cycle and small volume. This irreversible momentum urges the robot to switch from task to task efficiently. The biggest bottleneck here is the tedious programming, which also has high prerequisites for most practitioners in manufacturing. This situation motivates the development of a PBD system that can understand the assembly skills performed by the human experts in the demonstration and accordingly generate the program for robot's execution of the taught task. In this paper, we present a skill decoding system to parse the observational raw demonstration into symbolic sequences, which is the crucial bridge to enable the automatic programming. The system achieves high performance in recognition and is tailored for the PBD in assembly tasks by considering both advantages and disadvantages in the background of assembly, such as controllable environment and limited computational resources. It is particularly useful for assembly tasks with modularized actions based on a set of standard parts. At the perspective of industrial application, the PBD upon the proposed system is a promising solution to improve the flexibility of manufacture, which is expected to be true in midterm but an important step toward this goal.

Structured output support vector machine (SVM) based tracking algorithms have shown favorable performance recently. Nonetheless, the time-consuming candidate sampling and complex optimization limit their real-time applications. In this paper, we propose a novel large margin object tracking method which absorbs the strong discriminative ability from structured output SVM and speeds up by the correlation filter algorithm significantly. Secondly, a multimodal target detection technique is proposed to improve the target localization precision and prevent model drift introduced by similar objects or background noise. Thirdly, we exploit the feedback from high-confidence tracking results to avoid the model corruption problem. We implement two versions of the proposed tracker with the representations from both conventional hand-crafted and deep convolution neural networks (CNNs) based features to validate the strong compatibility of the algorithm. The experimental results demonstrate that the proposed tracker performs superiorly against several state-of-the-art algorithms on the challenging benchmark sequences while runs at speed in excess of 80 frames per second.

Many standard robotic platforms are equipped with at least a fixed 2D laser range finder and a monocular camera. Although those platforms do not have sensors for 3D depth sensing capability, knowledge of depth is an essential part in many robotics activities. Therefore, recently, there is an increasing interest in depth estimation using monocular images. As this task is inherently ambiguous, the data-driven estimated depth might be unreliable in robotics applications. In this paper, we have attempted to improve the precision of monocular depth estimation by introducing 2D planar observation from the remaining laser range finder without extra cost. Specifically, we construct a dense reference map from the sparse laser range data, redefining the depth estimation task as estimating the distance between the real and the reference depth. To solve the problem, we construct a novel residual of residual neural network, and tightly combine the classification and regression losses for continuous depth estimation. Experimental results suggest that our method achieves considerable promotion compared to the state-of-the-art methods on both NYUD2 and KITTI, validating the effectiveness of our method on leveraging the additional sensory information. We further demonstrate the potential usage of our method in obstacle avoidance where our methodology provides comprehensive depth information compared to the solution using monocular camera or 2D laser range finder alone.

Acquiring the accurate 3-D position of a target person around a robot provides fundamental and valuable information that is applicable to a wide range of robotic tasks, including home service, navigation and entertainment. This paper presents a real-time robotic 3-D human tracking system which combines a monocular camera with an ultrasonic sensor by the extended Kalman filter (EKF). The proposed system consists of three sub-modules: monocular camera sensor tracking model, ultrasonic sensor tracking model and multi-sensor fusion. An improved visual tracking algorithm is presented to provide partial location estimation (2-D). The algorithm is designed to overcome severe occlusions, scale variation, target missing and achieve robust re-detection. The scale accuracy is further enhanced by the estimated 3-D information. An ultrasonic sensor array is employed to provide the range information from the target person to the robot and Gaussian Process Regression is used for partial location estimation (2-D). EKF is adopted to sequentially process multiple, heterogeneous measurements arriving in an asynchronous order from the vision sensor and the ultrasonic sensor separately. In the experiments, the proposed tracking system is tested in both simulation platform and actual mobile robot for various indoor and outdoor scenes. The experimental results show the superior performance of the 3-D tracking system in terms of both the accuracy and robustness.

This paper presents a novel appearance and shape feature, RISAS, which is robust to viewpoint, illumination, scale and rotation variations. RISAS consists of a keypoint detector and a feature descriptor both of which utilise texture and geometric information present in the appearance and shape channels. A novel response function based on the surface normals is used in combination with the Harris corner detector for selecting keypoints in the scene. A strategy that uses the depth information for scale estimation and background elimination is proposed to select the neighbourhood around the keypoints in order to build precise invariant descriptors. Proposed descriptor relies on the ordering of both grayscale intensity and shape information in the neighbourhood. Comprehensive experiments which confirm the effectiveness of the proposed RGB-D feature when compared with CSHOT [1] and LOIND[2] are presented. Furthermore, we highlight the utility of incorporating texture and shape information in the design of both the detector and the descriptor by demonstrating the enhanced performance of CSHOT and LOIND when combined with RISAS detector.

In this paper, we develop a robust efficient visual SLAM system that utilizes heterogeneous point and line features. By leveraging ORB-SLAM [1], the proposed system consists of stereo matching, frame tracking, local mapping, loop detection, and bundle adjustment of both point and line features. In particular, as the main theoretical contributions of this paper, we, for the first time, employ the orthonormal representation as the minimal parameterization to model line features along with point features in visual SLAM and analytically derive the Jacobians of the re-projection errors with respect to the line parameters, which significantly improves the SLAM solution. The proposed SLAM has been extensively tested in both synthetic and real-world experiments whose results demonstrate that the proposed system outperforms the state-of-the-art methods in various scenarios.

Image representation has been intensively explored in the domain of computer vision for its significant influence on the relative tasks such as image clustering and classification. It is valuable to learn a low-dimensional representation of an image which preserves its inherent information from the original image space. At the perspective of manifold learning, this is implemented with the local invariant idea to capture the intrinsic low-dimensional manifold embedded in the high-dimensional input space. Inspired by the recent successes of deep architectures, we propose a local invariant deep nonlinear mapping algorithm, called graph regularized auto-encoder (GAE). With the graph regularization, the proposed method preserves the local connectivity from the original image space to the representation space, while the stacked auto-encoders provide explicit encoding model for fast inference and powerful expressive capacity for complex modeling. Theoretical analysis shows that the graph regularizer penalizes the weighted Frobenius norm of the Jacobian matrix of the encoder mapping, where the weight matrix captures the local property in the input space. Furthermore, the underlying effects on the hidden representation space are revealed, providing insightful explanation to the advantage of the proposed method. Finally, the experimental results on both clustering and classification tasks demonstrate the effectiveness of our GAE as well as the correctness of the proposed theoretical analysis, and it also suggests that GAE is a superior solution to the current deep representation learning techniques comparing with variant auto-encoders and existing local invariant methods.

In this paper, we present an online learning framework for traversable region detection fusing both appearance and geometry information. Our framework proposes an appearance classifier supervised by the sparse geometric clues to capture the variation in online data, yielding dense detection result in real time. It provides superior detection performance using appearance information with weak geometric prior and can be further improved with more geometry from external sensors. The learning process is divided into three steps: First, we construct features from the super-pixel level, which reduces the computational cost compared with the pixel level processing. Then we classify the multi-scale super-pixels to vote the label of each pixel. Second, we use weighted extreme learning machine as our classifier to deal with the imbalanced data distribution since the weak geometric prior only initializes the labels in a small region. Finally, we employ the online learning process so that our framework can be adaptive to the changing scenes. Experimental results on three different styles of image sequences, i.e., shadow road, rain sequence, and variational sequence, demonstrate the adaptability, stability, and parameter insensitivity of our weak geometry motivated method. We further demonstrate the performance of learning framework on additional five challenging data sets captured by Kinect V2 and stereo camera, validating the method's effectiveness and efficiency.

Scene classification is a fundamental perception task for environmental understanding in today's robotics. In this paper, we have attempted to exploit the use of popular machine learning technique of deep learning to enhance scene understanding, particularly in robotics applications. As scene images have larger diversity than the iconic object images, it is more challenging for deep learning methods to automatically learn features from scene images with less samples. Inspired by human scene understanding based on object knowledge, we address the problem of scene classification by encouraging deep neural networks to incorporate object-level information. This is implemented with a regularization of semantic segmentation. With only 5 thousand training images, as opposed to 2.5 million images, we show the proposed deep architecture achieves superior scene classification results to the state-of-the-art on a publicly available SUN RGB-D dataset. In addition, performance of semantic segmentation, the regularizer, also reaches a new record with refinement derived from predicted scene labels. Finally, we apply our model trained on SUN RGB-D dataset to a set of images captured in our university using a mobile robot, demonstrating the generalization ability of the proposed algorithm.

Autonomous robots enjoy a wide popularity nowadays and have been applied in many applications, such as home security, entertainment, delivery, navigation and guidance. It is vital for robots to track objects accurately in real time in these applications, so it is necessary to focus on tracking algorithms to improve the robustness, speed and accuracy. In this paper, we propose a real-time robust object tracking algorithm based on a hierarchical ensemble framework which incorporates information including individual pixel features, local patches and holistic target models. The framework combines multiple ensemble models simultaneously instead of using a single ensemble model individually. A discriminative model which accounts for the matching degree of local patches is adopted via a bottom ensemble layer, and a generative model which exploits holistic templates is used to search for the object based on the middle ensemble layer as well as an adaptive Kalman filter. We test the proposed tracker on challenging benchmark image sequences. The experimental results demonstrate that the proposed tracker performs superiorly against several state-of-the-art algorithms, especially when the appearance changes dramatically and the occlusions occur.

We introduce a novel RGB-D descriptor called local ordinal intensity and normal descriptor (LOIND) with the integration of texture information in RGB image and geometric information in depth image. We implement the descriptor with a 3-D histogram supported by orders of intensities and angles between normal vectors, in addition with the spatial sub-divisions. The former ordering information which is invariant under the transformation of illumination, scale and rotation provides the robustness of our descriptor, while the latter spatial distribution provides higher information capacity so that the discriminative performance is promoted. Comparable experiments with the state-of-art descriptors, e.g. SIFT, SURF, CSHOT and BRAND, show the effectiveness of our LOIND to the complex illumination changes and scale transformation. We also provide a new method to estimate the dominant orientation with only the geometric information, which can ensure the rotation invariance under extremely poor illumination.

In this paper, we present a novel learning framework for traversable region detection. Firstly, we construct features from the super-pixel level which can reduce the computational cost compared to pixel level. Multi-scale super-pixels are extracted to give consideration to both outline and detail information. Then we classify the multiple-scale super-pixels and merge the labels in pixel level. Meanwhile, we use weighted ELM as our classifier which can deal with the imbalanced class distribution since we only assume that a small region in front of robot is traversable at the beginning of learning. Finally, we employ the online learning process so that our framework can be adaptive to varied scenes. Experimental results on three different style of image sequences, i.e. shadow road, rain sequence and variational sequence, demonstrate the adaptability, stability and parameter insensitivity of our method to the varied scenes and complex illumination.

In robot programming by demonstration (PBD) for assembly tasks, one of the important topics is to inference the poses and spatial relations of parts during the demonstration. In this paper, we propose a world model called assembly graph (AG) to achieve this task. The model is able to represent the poses of all parts, the relations, observations provided by vision techniques and prior knowledge in a unified probabilistic graph. Then the problem is stated as likelihood maximization estimation of pose parameters with the relations being the latent variables. Classification expectation maximization algorithm (CEM) is employed to solve the model. Besides, the contradiction between relations is incorporated as prior knowledge to better shape the posterior, thus guiding the algorithm find a more accurate solution. In experiments, both simulated and real world datasets are applied to evaluate the performance of our proposed method. The experimental results show that the AG gives better accuracy than the relations as deterministic variables (RDV) employed in some previous works due to the robustness and global consistency. Finally, the solution is implemented into a PBD system with ABB industrial robotic arm simulator as the execution stage, succeeding in real world captured assembly tasks.

Feature selection tries to find a subset of feature from a larger feature pool and the selected subset can provide the same or even better performance compared with using the whole set. Feature selection is usually a critical preprocessing step for many machine-learning applications such as clustering and classification. In this paper, we focus on feature selection for supervised classification which targets at finding features that can best predict class labels. Traditional greedy search algorithms incrementally find features based on the relevance of candidate features and the class label. However, this may lead to suboptimal results when there are redundant features that may interfere with the selection. To solve this problem, we propose a subset selection algorithm that considers both the selected and remaining features' relevances with the label. The intuition is that features, which do not have better alternatives from the feature set, should be selected first. We formulate the selection problem as maximizing the dependency margin which is measured by the difference between the selected feature set performance and the remaining feature set performance. Extensive experiments on various data sets show the superiority of the proposed approach against traditional algorithms.

Temporal asynchrony between two cameras in the vision system is a usual problem in practice. In some vision task such as estimating fast moving targets, the estimation error caused by the tiny temporal asynchrony will become non-ignorable essentials. This paper will address on the asynchrony in the stereo vision system of humanoid Ping-Pong robot, and present a real-time accurate Ping-Pong ball trajectory estimation algorithm. In our approach, the complex Ping-Pong ball motion model is simplified by a polynomial parameter function of time t due to the limited observing time interval and the requirement of real-time computation. We then use the perspective projection camera model to re-project the ball's parameter function on time t into its image coordinates on both cameras. Based on the assumption that the time gap of two asynchronous cameras will maintain a const during very short time interval, we can obtain the time gap value and also the trajectory parameters of the Ping-Pong ball in a short time interval by minimizing the errors between the images of the ball in each camera and their re-projection images from the modeled parameter function on time t. Comprehensive experiments on real Ping-Pong robot cases are carried out, the results show our approach is more proper for the vision system of humanoid Ping-Pong robot, when concerning the accuracy and real-time performance simultaneously.

SUMMARY The mobile agent technique has been broadly used in next generation distributed systems. The system performance measurement and simulation are required before the system can be deployed on a large scale. In this paper, we address performance analysis on a finite state mobile agent prototype on the basis ofVirtual Hierarchical TreeGridOrganizations (VIRGO). The finite states refer to the migration, execution, and searching of the mobile agent. We introduce a novel evaluation model for the finite state mobile agent. The experimental results based on this evaluation model show that the finite mobile agents can perform well under multiple agent conditions and are superior to the traditional client/server approach. Copyright © 2013 John Wiley & Sons, Ltd.

This paper explains the mathematics of large scaled granular computing (GrC), augmented with a new Knowledge theory, by unifying rough set theories (RS) into one single concept, namely, neighborhood systems (NS). NS was first introduced in 1989 by T. Y. Lin to capture the concepts of “near” (topology) and “conflict” (security). Since 1996 when the term Granular Computing (GrC) was coined by T. Y. Lin to label Zadeh's vision, NS has been pushed into the “heart” of GrC. In 2011, LNS, the largest NS, was axiomatized; it implied that this set of axioms defines a new mathematics that realizes Zadeh's vision. The main messages are: this new mathematics is powerful and practical.

We propose granular computing (GrC) on ontology as a solution to the problem of modeling complex architectures. We expressed the architectures formally as ontology domains, which include two components: the set of basic vocabularies and a knowledge library of rules. The set of basic vocabularies contains elements or basic architecture components. The knowledge library comprises rules that control the combination and construction of the basic elements. As the rules are often given by architectural experts subjectively, they may contain redundant, conflicting, and overlapping rules, especially in certain styles of ancient southeast Chinese architecture. It is difficult to distinguish or identify these rules; therefore, we apply the multilevel approach on ontology [Y. Liu, C. Xu, Q. Zhang, and Y. Pan, ¿Smart architect: Scalable ontology-based modeling for ancient chinese architecture,¿ IEEE Intell. Syst., vol. 23, no. 1, pp. 49-56, Jan./Feb. 2008] and approximation theory of GrC. In this process, we present a measurement that is based on roughness functions to evaluate the degrees of approximation between the selected set and certain architecture domains. With the monotonicity characteristic of roughness functions, we can design a heuristic algorithm to select a suitable knowledge base (rule set) to assist in integrating the parts into final architectures, via several levels. Experiments with a real architectural project, i.e., modeling ancient southeast Chinese architectures, show that our method is effective and may simplify the design of the automodeling system and enhance its performance.

This paper presents a framework for Web service domain classification by granular computing. How to classify the Web services is a big issue because of fuzzy knowledge about these Web services. It is reasonable to assume that the Web services have vague boundaries of the domains they are belonged to. Here, we use roughness function measurement based on P-lower, NP-lower and P-upper,NP-upper approximations of P set (properties) to find the most suitable Web services. We also use the hierarchical algorithm to find the suitable Web service effective. The framework for Web service domain classification can be used in Web service discovery based on VIRGO domain classification P2P technologies.

An estimation distribution algorithm for path planning of an autonomous mobile robot is proposed in this paper. Our algorithm encodes each path to a fixed-length real number string, updates the populations by building probabilistic model of solution distributions. Simulations results show that the proposed algorithm works well in both static and dynamic environments.

Robot motion simulation is one of the challenging topic both in the character animation and the robotics. In this paper, a simulation system HumRoboSim is introduced. A 'real' humanoid robot is created in a virtual world considering the dynamics and collision in HumRoboSim. We present a new technique for robot motion generation with the robot dynamics model and constraints within the physical engine. Virtual sensors which provide accurate sensor data are also simulated to aid robot's attitude control and interaction with environment. We also propose a multi-agent control method which obtains the information from virtual sensors and makes decision to control the robot action. Our simulator could provide users with a convenient interface for manipulation and programming. Theexperimental results show that the proposed robot motion generation method can be successfully applied both in the virtual robot motion animation and the real hardware robot system. And the multiple agent control policy could be applied in similar hardware robot system.

Better information leads to better design. In landscape design, the designers care for what plans affect the urban more than what their plans look like. In this paper, we discuss some key techniques in a GIS based landscape design system. With the advantage of the GIS, designers can make better decisions. A prototype named landscape evaluation system (LES) has been implemented. In LES system, designers can build up their virtual evaluation world, and establish the interrelationship of the entities in their virtual world.