Incomplete cross-modal retrieval (ICMR) requires models to recover missing modalities and robustly align heterogeneous ones for effective retrieval. Existing methods, however, fall short in both aspects. They often rely on limited semantic cues, such as single samples or coarse category prototypes, which compromises reconstruction quality. Moreover, these approaches are vulnerable to learning spurious cross-modal correlations, thereby impairing accurate alignment and hindering retrieval performance. To address these challenges, we propose Causality-Aligned Semantic Recovery (CASR), a novel method designed to both comprehensively restore missing modalities and mitigate spurious associations between vision and language. Our CASR involves two essential components: i) the Missing Modality Imagination (MMI) module, which combines category semantic priors with relevant contextual information to achieve high-quality semantic reconstruction; ii) the Explicit Causal Alignment (ECA) module, which explicitly learns environment-invariant attention, effectively eliminating the interference of spurious correlations and improving retrieval performance. Furthermore, we extend CASR to the challenging task of Partially Aligned Cross-Modal Retrieval, where we treat unlabeled unpaired data as a form of incomplete data. By leveraging MMI and ECA modules, we are able to learn robust representations in this setting. Extensive experiments on benchmark datasets under various missing rates demonstrate that CASR achieves superior robustness and retrieval performance.

Universal Cross-Domain Retrieval (UCDR) aims to retrieve images across unseen domains and categories, a critical capability for real-world applications. While large-scale Vision-Language Models (VLMs) like CLIP offer strong zero-shot category generalization, they struggle with domain shifts. Existing methods often improve domain robustness at the cost of high computational overhead or by compromising the VLM's inherent knowledge. To address this, we propose Decoupled and Fused Tuning with LoRA (DeFT-LoRA), a novel and parameter-efficient framework that integrates Low-Rank Adaptation (LoRA) with a Mixture-of-Experts (MoE) mechanism. This approach resolves the intrinsic conflict between domain-invariant and domain-specific knowledge in a single adapter, enabling our model to construct a domain adapters for each input image. We propose a three-stage training strategy, which first learns a shared Base LoRA for domain-invariant features, then derives Domain-Specific Experts to capture specific styles, and finally fuses them dynamically with a lightweight gating network. Extensive experiments on three UCDR benchmarks demonstrate that DeFT-LoRA achieves comparable or superior performance to state-of-the-art methods while requiring only 1.46 percent of CLIP's image-encoder parameters and reducing computational overhead, thereby establishing an exceptional balance between accuracy and efficiency.

Amidst a shortage of qualified mental health professionals, the integration of large language models (LLMs) into psychological applications offers a promising way to alleviate the growing burden of mental health disorders.Recent reasoning-augmented LLMs have achieved remarkable performance in mathematics and programming, while research in the psychological domain has predominantly emphasized emotional support and empathetic dialogue, with limited attention to reasoning mechanisms that are beneficial to generating accurate responses.Therefore, in this paper, we propose Psyche-R1, the first Chinese psychological LLM that jointly integrates empathy, psychological expertise, and reasoning, built upon a novel data curation pipeline.Specifically, we design a comprehensive data synthesis pipeline that produces over 75k high-quality psychological questions paired with detailed rationales, generated through an iterative prompt-rationale optimization procedure, along with 73k empathetic dialogues.Subsequently, we employ a hybrid training strategy wherein challenging samples are identified through a multi-LLM cross-selection strategy for group relative policy optimization (GRPO) to improve reasoning ability, while the remaining data are used for supervised fine-tuning (SFT) to enhance empathetic response generation and psychological domain knowledge.Extensive experiment results demonstrate the effectiveness of Psyche-R1 across several psychological benchmarks, where our 7B Psyche-R1 achieves comparable results to 671B DeepSeek-R1.

Naixin Zhai, Pengyang Shao, Binbin Zheng, Yonghui Yang, Fei Shen, Long Bai, Xun Yang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Cognitive Diagnosis (CD) models are constrained by the data quality of students' response logs. Recent advancements in Large Language Model (LLM) based data augmentation show promise for enhancing CD. However, ensuring the reliability and accuracy of LLM-generated annotations remains a significant challenge. In this paper, we propose Multi-Agent based Concept Augmentation for Cognitive Diagnosis (MACA-CD), a novel approach that enhances CD by generating and fusing reliable concept descriptions and relations based solely on concept names. MACA-CD consists of two main components: (1) a Multi-Agent Debate (MAD) based concept augmentation process that generates diverse and reliable concept descriptions and relations, reducing reliance on behavioral data. For concept descriptions, two agents generate outputs that include definitions, core features, and real-world applications, and continue debating until a judge agent determines that consensus has been reached. Concept relations are then identified using a Breadth-First Search approach to efficiently and progressively uncover relationships based on concept descriptions, with each step carried out by MAD. (2) a concept augmentation-enhanced CD model that refines concept embeddings using a graph self-supervised learning fusion layer and a pairwise comparator-based Description Fusion Layer, leading to more reliable and accurate concept embeddings. Experimental results on three real-world datasets show that MACA-CD consistently outperforms existing methods under various real-world scenarios.

As Large Language Models (LLMs) increasingly shape online content, how to remove targeted information from well-trained LLMs (also known as LLM unlearning) has become increasingly critical for web governance. A key challenge in LLM unlearning lies in the sample-wise imbalance within the forget set: different samples exhibit widely varying unlearning difficulty, leading to asynchronous forgetting speeds where some knowledge remains insufficiently erased while others become over-forgotten. To address this challenge, we propose BalDRO, a novel and efficient framework for balanced LLM unlearning. BalDRO formulates unlearning as a min–sup process, where the inner process identifies a worst-case data distribution that adaptively emphasizes hard-to-unlearn samples, while the outer process updates model parameters based on the worst-case data distribution. We instantiate this formulation through two efficient variants: BalDRO-G, a discrete GroupDRO-based approximation that focuses on high-loss subsets, and BalDRO-DV, a continuous Donsker–Varadhan dual method that enables smooth, adaptive weighting within standard LLM training pipelines. Extensive experiments on the TOFU and MUSE benchmarks demonstrate the effectiveness of our proposed BalDRO, yielding significant improvements in both forgetting quality and model utility over existing methods. For reproducibility, we have released the code for BalDRO. https://github.com/nxZhai/BalDRO.

Depressive disorders represent a major global public health challenge. As an increasing number of individuals share their emotional experiences and concerns on social media, researchers have shown growing interest in leveraging such data for early depression screening. However, most existing methods rely on a fixed model and a singular reasoning paradigm, which constrains their adaptability to depression detection. The limited availability of mental health-related data and variability in training data distributions across different LLMs hinder their consistent and comprehensive understanding of diverse psychological symptoms. In this paper, we propose AdaDepression, a framework that enables explainable depression screening through a two-hop retrieval algorithm to identify symptom-relevant posts and a two-stage adaptive routing mechanism for selecting appropriate reasoning strategies and LLMs. Specifically, we first collect representative posts from the training dataset to capture the real-world symptom expressions, and then utilize these posts to retrieve symptom-relevant posts from the user's posting history. Subsequently, we employ the Mixture of Routers (MoR), which integrates the Mixture of Experts (MoE) into the routing mechanism to select the optimal reasoning strategies and LLMs in a cascaded manner. Finally, we complete the standardized psychological questionnaire using the selected LLMs and reasoning strategies. Experimental results on the Reddit-based benchmarks demonstrate the effectiveness of the proposed method, outperforming existing studies on various metrics. Our code is released at https://github.com/MindIntLab-HFUT/AdaDepression.

In current text-to-video retrieval (T2VR), videos to be retrieved have been properly trimmed so that a correspondence between the videos and ad-hoc textual queries naturally exists. Note in practice that videos circulated on the Internet and social media platforms, while being relatively short, are typically rich in their content. Often, multiple scenes / actions / events are shown in a single video, leading to a more challenging T2VR setting wherein only part of the video content is relevant w.r.t. a given query. This paper presents a first study on this setting which we term Partially Relevant Video Retrieval (PRVR). Considering that a video typically consists of multiple moments, a video is regarded as partially relevant w.r.t. to a given query if it contains a query-related moment. We formulate the PRVR task as a multiple instance learning problem, and propose a Multi-Scale Similarity Learning (MS-SL++) network that jointly learns both clip-scale and frame-scale similarities to determine the partial relevance between video-query pairs. Extensive experiments on three diverse video-text datasets (TVshow Retrieval, ActivityNet-Captions and Charades-STA) demonstrate the viability of the proposed method.

One-shot Federated Learning (OFL) has emerged as a promising paradigm, enabling global model training with minimal communication overhead. In OFL, the server model is usually distilled from an ensemble of pre-trained client models, while the ensemble also facilitates synthetic data generation for the knowledge distillation process. Prior works show that the performance of the final model is fundamentally tied to both the quality of the synthetic data and the ensemble. However, existing methods often optimize these two components separately, overlooking their interaction. To address this coupled optimization problem and provide a unified solution to the dual challenges of data and model heterogeneity inherent in OFL, we introduce Co-Boosting++, a novel OFL framework where synthetic data generation and ensemble construction mutually enhance each other in an iterative fashion. First, we fix the ensemble and generate hard samples in an adversarial manner. These samples are crucial for enhancing the robustness of knowledge transfer, as they challenge the model to generalize better, thereby improving quality of the synthetic data and subsequent distillation process. Second, leveraging these hard samples, we enhance the ensemble via a Mixture of Experts (MoE) mechanism. MoE allows dynamic adjustment of ensemble weights based on the generated hard samples, which enables the ensemble to better capture diverse and heterogeneous knowledge from client models. Furthermore, we extend Co-Boosting++ to support the simultaneous generation of multiple heterogeneous target models, enabling efficient adaptation to diverse device constraints. Extensive experiments on benchmark datasets demonstrate that Co-Boosting++ consistently outperforms state-of-the-art methods due to its coupled optimization of data and ensemble quality. Additionally, Co-Boosting++ is highly practical in real-world model market scenarios, requiring no local training modifications, additional transmissions, or restrictions on client model architectures.

Emotional support conversation systems have emerged as a promising complement to traditional mental health consultations, offering context-aware dialogue to support seekers’ emotional well-being. Despite their potential, two fundamental challenges remain unresolved: 1) modeling long-term emotional trajectory beyond short-term relief; and 2) adapting support strategies to context in a psychologically coherent manner. To address these challenges, we propose CAIWO, a novel framework that integrates world modeling and causality-enhanced imitation learning to systematically support seekers, alleviate psychological stress, and restore emotional balance. Specifically, CAIWO comprises two core components. The first is an emotional world model, which captures long-term emotional trajectories from historical interactions to inform anticipatory guidance. The second is a causality-enhanced imitation learning module, which infers latent causal dependencies to facilitate coherent strategy transitions and mitigate compounding errors typical of conventional imitation learning. By incorporating the final latent variables into the response decoder, CAIWO dynamically adjusts the strategies and generates emotionally resonant responses. Extensive experiments on the ESConv benchmark demonstrate that CAIWO outperforms state-of-the-art baselines by 8.6%, significantly improving the generation of responses that align with seekers’ emotional development and psychological needs.

Depression detection based on social media content has received increasing attention in recent years, as it allows for early diagnosis before the user’s psychological state deteriorates. Although traditional methods of depression detection can provide a classification of whether the user is depressed or not, they cannot provide human-like explanations and interactions. In this article, we propose a next-generation paradigm for depression detection, namely an interpretable and interactive depression detection system based on large language models (LLMs). The proposed system not only yields a final diagnosis result, but also offers diagnostic evidence grounded in established diagnostic criteria. Furthermore, it enables users to engage in natural language dialogue with the system, facilitating a more personalized understanding of their mental state based on their social media content. The interactive dialogue allows for the provision of tailored recommendations, which users can utilize to enhance their well-being. In constructing the entire system, we also addressed some nontrivial challenges. First, we introduced the chain of thoughts technique and professional depression diagnostic criteria when constructing the prompts, enabling our system to make decisions based on professional diagnosis criteria and provide explanations. Second, LLMs are incapable of processing excessively long contextual texts, and the accumulated posts of a single user may amount to tens of thousands of words. To overcome this limitation, we integrated a tweet selector that selects the part of posts for diagnosis. The experiments demonstrate that our depression detection system achieves the best performance across various settings, including full data setting, few-shot setting, zero-shot setting, independent-identical-distribution (IID) setting, and out-of-distribution (OOD) setting. Additionally, case studies reveal the explanation and interactivity of our system.

Low-Rank Adaptation (LoRA) is a widely utilized technique in topic of Parameter-Efficient Transfer Learning (PETL) which could use a limited number of trainable parameters to adapt the model to various downstream tasks. However, the setting of the locations and low-rank sizes in traditional LoRA relies heavily on the fixed and empirical values, which may hinder adaptability and lead to sharply decreasing performance, especially on some self-supervised pre-trained models. To alleviate this dilemma, we introduce a feature responsive LoRA (ResLoRA) method, a resource-efficient algorithm that automatically determines the LoRA modules’ required size based on the downstream task’s response. Firstly, we propose a Feature Decomposition loss (FD-loss) which leverages the feature singular values to mine the corresponding features of different downstream tasks, making the model parameters able to adequately represent downstream tasks. Subsequently, we leverage the Taylor expansion to measure the salience of the model parameters, then some high-efficient parameters with high significance could be leveraged to design a dynamically responsive LoRA. Specifically, the location and low-rank sizes of LoRA are determined based on the response parameters of the features for downstream tasks. Extensive experiments show that our ResLoRA achieves state-of-the-art performance, especially in the transfer capability of self-supervised models based on MoCo v3 and MAE. Our code is available at: https://github.com/wildboarman/ResLoRA.

Based on its excellent capability to extract temporal features, transformer has been widely used in monocular 3D human pose estimation. However, due to its global perspective, it performs inadequately in extracting spatial features, which hinders breakthroughs in performance. In this paper, we propose a local-global feature fusion method based on GCN and transformer for 3D human pose estimation. Our method integrates GCN with multiscale transformer to extract local spatiotemporal features of poses. These are then integrated with the global spatiotemporal features extracted by vanilla transformer to reconstruct 3D human poses accurately. In addition, we introduce a hierarchical feature fusion method to better capturing the underlying 3D pose structure. It blends deep abstract features with shallow raw features. We evaluate our model on the Human3.6M and MPI-INF-3DHP datasets, and experimental results demonstrate that our approach outperforms existing state-of-the-art methods. We achieve advanced performance on both datasets with errors of 37.7mm and 16.4mm under MPJPE, respectively. The code and model are available at https://github.com/ygx7/LG3DPose.

Recently, the cross-modal retrieval task has gained significant attention with the advent of large-scale vision-language pretraining models, e.g., CLIP. These methods typically map the vision and language modalities into a shared embedding space and then build similarity relations based on the joint feature representations. Despite tremendous progress in this field, most existing methods still suffer from unreliable retrieval results caused by data and model uncertainties, which can arise from inherent data ambiguity or noisy pairs. In this article, we propose a novel cross-modal retrieval framework with aleatoric-epistemic joint uncertainty modeling (AEUM). AEUM is committed to providing reliable uncertainty estimation for both data (aleatoric uncertainty, AU) and model (epistemic uncertainty, EU), which are then used to correct the initial cross-modal similarity to yield more accurate retrieval results. Specifically, for AU, we introduce learnable semantic tokens for each modality to estimate the data-induced uncertainty in another modality, offering guidance on data complexity or ambiguity. For the EU, we leverage the efficient evidential learning paradigm to estimate model-induced uncertainty and incorporate it into the model's predictions, thereby enhancing robustness against noisy data. Extensive experiments demonstrate the effectiveness and generalization of our method on multiple cross-modal retrieval benchmarks, including five video-text retrieval datasets (MSRVTT, LSMDC, MSVD, VATEX, and DiDeMo) and two image-text retrieval datasets (MSCOCO and Flickr30K). Our code is publicly available at https://github.com/cty8998/AEUM.

Affective Explanation Captioning (AEC) aims to perform viewer-centered visual emotion analysis by not only identifying the emotions evoked by an image but also explaining their underlying causes. Prior efforts have achieved promising results by fine-tuning LLMs on affective data; however, two key challenges remain: 1) the inherent subjectivity of human emotion leads to diverse interpretations of the same image, making it difficult for models to catch dominant emotions; and 2) the affective gap between abstract emotions and concrete visual content hinders models from capturing both semantic and emotional aspects effectively. To tackle these challenges, we propose Consensus-Prompted Emotion Reasoning (CPER), a new framework that explicitly models emotional diversity and enforces emotional-semantic alignment. Inspired by psychological studies, we observe that common emotional patterns often emerge within certain groups, which we refer to as affective consensus. Capturing this consensus across varying levels is helpful for bridging the subjectivity in AEC. Specifically, we introduce a consensus-based bucket prompt, which depicts the consensus level of each emotional perspective, serving as a control signal to adjust the emotion reasoning. To reconcile abstract emotion understanding and concrete visual grounding, we design a dual-space representation, where a CLIP encoder extracts objective semantic evidence and an emotion encoder captures abstract affective cues for AEC. Furthermore, an emotion consistency learning strategy is devised, which explicitly aligns the generated explanation with the input image and the emotion label, ensuring both emotionally and semantically grounded explanations. Extensive experiments on three benchmark datasets, ranging from visual arts (ArtEmis v1.0 and ArtEmis v2.0) and real-world images (Affection), demonstrate the effectiveness of our CPER in terms of emotional diversity and semantic coherence compared to state-of-the-art methods. Our code is publicly available at https://github.com/songpipi/CPER.

Test-time adaptation (TTA) has emerged as a key strategy to enhance vision-language models (VLMs) under real-world distribution changes. However, existing methods always face two problems: 1) The fundamental trade-off dilemma: parameter-free TTA retains inference efficiency but fails to correct modality misalignment, while prompt tuning adapts to shifts, incurs high computational costs, and lacks knowledge retention. 2) Discriminative collapse also exists in TTA when faced with fine-grained downstream tasks. To alleviate these two bottlenecks, we introduce Style-aware Contrastive Test-Time Adaptation (SCTTA), a novel framework that jointly addresses modality misalignment and discriminative collapse. Firstly, we introduce Style-aware Embedding Adaptation (SEA), which dynamically refines text embeddings by incorporating domain-specific style attributes, improving alignment between visual and textual modalities. Secondly, we propose Fine-grained Contrastive Adaptation (FCA), which enhances feature separation by enforcing contrastive learning with adaptive prototypes, reducing inter-class feature overlap in fine-grained tasks. In addition, we introduce Dual-Cache Model (DCM), which extends prior unimodal cache model to a multimodal cache for the first time. Eventually, it accumulates adaptation knowledge through a visual-cache (capturing evolving domain styles) and a textual-cache (retaining discriminative semantics), enabling long-term adaptation without additional overhead. Extensive experiments on 15 datasets demonstrate that our approach achieves state-of-the-art performance for both fine-grained and out-of-distribution dataset benchmarks. Furthermore, SCTTA continuously improves as more test samples accumulate, validating its sustainable adaptation capacity. Our code is available at https://github.com/alusi123/SCTTA.

Knowledge Distillation (KD) is a widely popular model compression technique that can effectively transfer knowledge from a pre-trained, large-scale teacher model to a more compact and lightweight student model. Traditional KD methods aim to improve the student’s representation capability by mimicking the teacher’s features, e.g., minimizing the \(\mathcal{L}_{2}\) distance between their intermediate features. However, due to the capacity gap between the student and the teacher, student often struggles to precisely mimic the features of the teacher. To address this challenge, we propose to boost the knowledge distillation for the visual recognition tasks via Direction Alignment and Discrepancy Standardization ( DADS) , which exploits the feature scaling technique to distill from both the feature direction and feature discrepancy. To this end, we devise an efficient feature alignment module to align the dimensions of teacher and student features. Moreover, we align the direction of student features and teacher features, which are pre-processed by normalization. Furthermore, we leverage the Kullback–Leibler (KL) divergence to refine the features alignment, minimizing discrepancy in the distribution of features across samples, which is pre-processed by \(\mathcal{Z}\) -score standardization. In this way, our proposed approach can effectively transfer the knowledge from the teacher to the student, facilitating the downstream visual recognition applications, such as image classification and semantic segmentation. Extensive experimental analyses clearly validate the effectiveness of DADS . Compared with previous KD methods, our approach sets a new benchmark, achieving state-of-the-art results on visual recognition tasks.

In the realm of video dialog response generation, capturing both the essence of video content and the temporal nuances of conversation history is crucial. While some approaches rely on large-scale pretrained visual-language models, often neglecting temporal dynamics, others emphasize spatial-temporal relationships within videos but demand intricate object trajectory pre-extractions and overlook dialog temporal dynamics. This paper introduces the Dual Temporal Grounding-enhanced Video Dialog model (DTGVD), designed to bridge the gap between these two approaches. DTGVD uniquely integrates the strengths of both by emphasizing dual temporal relationships. It achieves this by predicting dialog turn-specific temporal regions, selectively filtering video content, and grounding responses in both video and dialog contexts.A key innovation of DTGVD is its advanced handling of chronological interplay within dialogs. By effectively capturing and leveraging dependencies between dialog turns, it enables a more nuanced understanding of conversational dynamics. To further align video and dialog temporal dynamics, we introduce a list-wise contrastive learning strategy. In this framework, accurately grounded turn-clip pairings are treated as positive samples, while less precise pairings serve as negative samples. This refined classification is then seamlessly integrated into our end-to-end response generation mechanism.Evaluations using AVSD@DSTC-7 and AVSD@DSTC-8 datasets underscore the superiority of our methodology.

Existing efforts in text-based video question answering (TextVideoQA) are criticized for their opaque decision-making and heavy reliance on scene-text recognition. In this paper, we propose to studyGrounded TextVideoQAby forcing models to answer questions and spatio-temporally localize the relevant scene-text regions, thus decoupling QA from scene-text recognition and promoting research towards interpretable QA. The task has three-fold significance. First, it encourages scene-text evidence versus other short-cuts for answer predictions. Second, it directly accepts scene-text regions as visual answers, thus circumventing the problem of ineffective answer evaluation by stringent string matching. Third, it isolates the challenges inherited in VideoQA and scene-text recognition. This enables the diagnosis of the root causes for failure predictions,e.g., wrong QA or wrong scene-text recognition? To achieve Grounded TextVideoQA, we propose the T2S-QA model that highlights a disentangled temporal-to-spatial contrastive learning strategy for weakly-supervised scene-text grounding and grounded TextVideoQA. To facilitate evaluation, we construct a new datasetViTXT-GQAwhich features 52Kscene-text bounding boxes within 2.2Ktemporal segments related to 2Kquestions and 729 videos. With ViTXT-GQA, we perform extensive experiments and demonstrate the severe limitations of existing techniques in Grounded TextVideoQA. While T2S-QA achieves superior results, the large performance gap with human leaves ample space for improvement. Our further analysis of oracle scene-text inputs posits that the major challenge is scene-text recognition. To advance the research of Grounded TextVideoQA, our dataset and code are athttps://github.com/zhousheng97/ViTXT-GQA.git

Cognitive Diagnosis Modeling is a fundamental task in intelligent education, intending to assess students’ mastery levels on knowledge concepts through interactions. Previous methodologies prioritized enhancing average diagnostic accuracy. However, they often neglected the Long-Tailed issue in interactions. To address this issue, we propose an Adaptive Self-Supervised Graph Learning for Cognitive Diagnosis (ASCD) framework that leverages relatively balanced sparse views to forcing the graph network to focus on long-tailed nodes, aiming to tackle the long-tailed problem in graph-based cognitive diagnosis. Additionally, we employ self-supervised manners to mitigate the impact of dropped information on other nodes. Our approach leverages the adaptive graph confusion method to create sparse views of the original student-exercise interaction graph. In these sparse views, both long-tailed and head students carry similar weights, pushing the graph network to allocate attention impartially across all students. We integrate two different graph confusion techniques into adaptive graph confusion to accommodate varying degrees of data sparsity: edge dropout and feature masking. ASCD can serve as a plug-and-play module integrated into any graph-based cognitive diagnosis model, enhancing its performance regarding long-tailed scenarios. Extensive experiments on real-world datasets show the effectiveness of our approach, especially on the students with much sparser interaction records.

Creating high-fidelity, coherent long videos is a sought-after aspiration. While recent video diffusion models have shown promising potential, they still grapple with spatiotemporal inconsistencies and high computational resource demands. We propose Global-Local Collaborative Diffusion (GLC-Diffusion), a tuning-free method for long video generation. It models the long video denoising process by establishing denoising trajectories through Global-Local Collaborative Denoising (GLCD) to ensure overall content consistency and temporal coherence between frames. Additionally, we introduce a Noise Reinitialization strategy which combines local noise shuffling with frequency fusion to improve global content consistency and visual diversity. Further, we propose a Video Motion Consistency Refinement (VMCR) module that computes the gradient of pixel-wise and frequency-wise losses to enhance visual consistency and temporal smoothness. Extensive experiments, including quantitative and qualitative evaluations on videos of varying lengths (e.g., 3× and 6× longer), demonstrate that our method effectively integrates with existing video diffusion models, producing coherent, high-fidelity long videos superior to previous approaches.

3D visual grounding (3DVG), which aims to correlate a natural language description with the target object within a 3D scene, is a significant yet challenging task. Despite recent advancements in this domain, existing approaches commonly encounter a shortage: a limited amount and diversity of text-3D pairs available for training. Moreover, they fall short in effectively leveraging different contextual clues (e.g., rich spatial relations within the 3D visual space) for grounding. To address these limitations, we propose AugRefer, a novel approach for advancing 3D visual grounding. AugRefer introduces cross-modal augmentation designed to extensively generate diverse text-3D pairs by placing objects into 3D scenes and creating accurate and semantically rich descriptions using foundation models. Notably, the resulting pairs can be utilized by any existing 3DVG methods for enriching their training data. Besides, AugRefer presents a language-spatial adaptive decoder that effectively adapts the potential referring objects based on the language description and various 3D spatial relations. Extensive experiments on three benchmark datasets clearly validate the effectiveness of AugRefer.

Recent methods using diffusion models have made significant progress in human image generation with various control signals such as pose priors. However, existing efforts are still struggling to generate high-quality images with consistent pose alignment, resulting in unsatisfactory output. In this paper, we propose a framework that delves into the graph relations of pose priors to provide control information for human image generation. The main idea is to establish a graph topological structure between the pose priors and latent representation of diffusion models to capture the intrinsic associations between different pose parts. A Progressive Graph Integrator (PGI) is designed to learn the spatial relationships of the pose priors with the graph structure, adopting a hierarchical strategy within an Adapter to gradually propagate information across different pose parts. Besides, a pose perception loss is introduced based on a pretrained pose estimation network to minimize the pose differences. Extensive qualitative and quantitative experiments conducted on the Human-Art and LAION-Human datasets clearly demonstrate that our model can achieve significant performance improvement over the latest benchmark models.

The goal of point cloud localization based on linguistic description is to identify a 3D position using textual description in large urban environments, which has potential applications in various fields, such as determining the location for vehicle pickup or goods delivery. Ideally, for a textual description and its corresponding 3D location, the objects around the 3D location should be fully described in the text description. However, in practical scenarios, e.g., vehicle pickup, passengers usually describe only the part of the most significant and nearby surroundings instead of the entire environment. In response to this partially relevant challenge, we propose CMMLoc, an uncertainty-aware Cauchy-Mixture-Model (CMM) based framework for text-to-point-cloud Localization. To model the uncertain semantic relations between text and point cloud, we integrate CMM constraints as a prior during the interaction between the two modalities. We further design a spatial consolidation scheme to enable adaptive aggregation of different 3D objects with varying receptive fields. To achieve precise localization, we propose a cardinal direction integration module alongside a modality pre-alignment strategy, helping capture the spatial relationships among objects and bringing the 3D objects closer to the text modality. Comprehensive experiments validate that CMMLoc outperforms existing methods, achieving state-of-the-art results on the KITTI360Pose dataset. Codes are available in this GitHub repository https://github.com/kevin301342/CMMLoc.

We introduce EgoTextVQA, a novel and rigorously constructed benchmark for egocentric QA assistance involving scene text. EgoTextVQA contains 1.5K ego-view videos and 7K scene-text aware questions that reflect real user needs in outdoor driving and indoor house-keeping activities. The questions are designed to elicit identification and reasoning on scene text in an egocentric and dynamic environment. With EgoTextVQA, we comprehensively evaluate 10 prominent multimodal large language models. Currently, all models struggle, and the best results (Gemini 1.5 Pro) are around 33% accuracy, highlighting the severe deficiency of these techniques in egocentric QA assistance. Our further investigations suggest that precise temporal grounding and multi-frame reasoning, along with high resolution and auxiliary scene-text inputs, are key for better performance. With thorough analyses and heuristic suggestions, we hope EgoTextVQA can serve as a solid testbed for research in egocentric scene-text QA assistance. Our dataset is released at: https://github.com/zhousheng97/EgoTextVQA.

The development of Emotional Support Conversation (ESC) systems is critical for delivering mental health support tailored to the needs of help-seekers.Recent advances in large language models (LLMs) have contributed to progress in this domain, while most existing studies focus on generating responses directly and overlook the integration of domain-specific reasoning and expert interaction.Therefore, in this paper, we propose a training-free Multi-Agent collaboration framework for ESC (Mul-tiAgentESC).The framework is designed to emulate the human-like process of providing emotional support through three stages: dialogue analysis, strategy deliberation, and response generation.At each stage, a multi-agent system is employed to iteratively enhance information understanding and reasoning, simulating real-world decision-making processes by incorporating diverse interactions among these expert agents.Additionally, we introduce a novel response-centered approach to handle the one-to-many problem on strategy selection, where multiple valid strategies are initially employed to generate diverse responses, followed by the selection of the optimal response through multi-agent collaboration.Experiments on the ESConv dataset reveal that our proposed framework excels at providing emotional support as well as diversifying support strategy selection 1 .

Open-vocabulary video visual relation detection (VidVRD) expands the scope of detecting object relations in videos to include unseen categories. It marks considerable advancement in recognizing novel relations solely by training on a base set, thus extending the frontiers of automated video understanding. However, the performance of current methods on novel predicates remains significantly inferior to that on base categories. We attribute this discrepancy to two primary factors: (1) A significant task misalignment between the Visual Relation Detection (VRD) task and the pre-trained models’ visual feature extractors, which are often designed for tasks like video-text retrieval and image-text retrieval, resulting in poor generalization to the novel set. (2) The relatively small size and limited vocabulary of open-vocabulary datasets, which create a substantial gap between base and novel predicates. Consequently, text prompts trained on the base set fail to generalize effectively to the novel set. To address these issues, we propose two improvement measures: (1) We decompose base and novel relations into actional and spatial patterns and introduce an innovative text prompt learning method that leverages the shared patterns between base and novel relations. (2) We develop a relation probability adjustment mechanism that utilizes reliable base relation predictions to adjust the probabilities of relations in novel classes by considering their overlaps in either actional or spatial contents. Experimental results on the benchmark dataset demonstrate significant performance improvements.

Existing text-to-image models often rely on parameter fine-tuning techniques such as Low-Rank Adaptation (LoRA) to customize visual attributes. However, when combining multiple LoRA models for content-style fusion tasks, unstructured modifications of weight matrices often lead to undesired feature entanglement between content and style attributes. We propose QR-LoRA, a novel fine-tuning framework leveraging QR decomposition for structured parameter updates that effectively separate visual attributes. Our key insight is that the orthogonal Q matrix naturally minimizes interference between different visual features, while the upper triangular R matrix efficiently encodes attribute-specific transformations. Our approach fixes both Q and R matrices while only training an additional task-specific $ΔR$ matrix. This structured design reduces trainable parameters to half of conventional LoRA methods and supports effective merging of multiple adaptations without cross-contamination due to the strong disentanglement properties between $ΔR$ matrices. Experiments demonstrate that QR-LoRA achieves superior disentanglement in content-style fusion tasks, establishing a new paradigm for parameter-efficient, disentangled fine-tuning in generative models. The project page is available at: https://luna-ai-lab.github.io/QR-LoRA/.

Knowledge editing aims to efficiently and cost-effectively correct inaccuracies and update outdated information. Recently, there has been growing interest in extending knowledge editing from Large Language Models (LLMs) to Multimodal Large Language Models (MLLMs), which integrate both textual and visual information, introducing additional editing complexities. Existing multimodal knowledge editing works primarily focus on text-oriented, coarse-grained scenarios, failing to address the unique challenges posed by multimodal contexts. In this paper, we propose a visual-oriented, fine-grained multimodal knowledge editing task that targets precise editing in images with multiple interacting entities. We introduce the Fine-Grained Visual Knowledge Editing (FGVEdit) benchmark to evaluate this task. Moreover, we propose a Multimodal Scope Classifier-based Knowledge Editor (MSCKE) framework. MSCKE leverages a multimodal scope classifier that integrates both visual and textual information to accurately identify and update knowledge related to specific entities within images. This approach ensures precise editing while preserving irrelevant information, overcoming the limitations of traditional text-only editing methods. Extensive experiments on the FGVEdit benchmark demonstrate that MSCKE outperforms existing methods, showcasing its effectiveness in solving the complex challenges of multimodal knowledge editing.

Knowledge editing aims to update outdated information in Large Language Models (LLMs). A representative line of study is locate-then-edit methods, which typically employ causal tracing to identify the modules responsible for recalling factual knowledge about entities. However, we find these methods are often sensitive only to changes in the subject entity, leaving them less effective at adapting to changes in relations. This limitation results in poor editing locality, which can lead to the persistence of irrelevant or inaccurate facts, ultimately compromising the reliability of LLMs. We believe this issue arises from the insufficient precision of knowledge localization. To address this, we propose a Fine-grained Neuron-level Knowledge Editing (FiNE) method that enhances editing locality without affecting overall success rates. By precisely identifying and modifying specific neurons within feed-forward networks, FiNE significantly improves knowledge localization and editing. Quantitative experiments demonstrate that FiNE efficiently achieves better overall performance compared to existing techniques, providing new insights into the localization and modification of knowledge within LLMs.

Remote sensing image-text retrieval is a fundamental task in remote sensing multimodal analysis, promoting the alignment of visual and language representations. The mainstream approaches commonly focus on capturing shared semantic representations between visual and textual modalities. However, the inherent characteristics of remote sensing image-text pairs lead to a semantic confusion problem, stemming from redundant visual representations and high inter-class similarity. To tackle this problem, we propose a novel Discriminative and Fine-grained Information Mining (DFIM) model, which aims to enhance semantic clarity by reducing visual redundancy and increasing the semantic gap between different classes. Specifically, the Dynamic Visual Enhancement (DVE) module adaptively enhances the visual discriminative features under the guidance of multimodal fusion information. Meanwhile, the Fine-grained Semantic Matching (FSM) module cleverly models the matching relationship between image regions and text words as an optimal transport problem, thereby refining intra-instance matching. Extensive experiments on two benchmark datasets justify the superiority of DFIM in terms of retrieval accuracy and visual interpretability over the leading methods.

Despite their strong performance in multimodal emotion reasoning, existing Multimodal Large Language Models (MLLMs) often overlook the scenarios involving emotion conflicts, where emotional cues from different modalities are inconsistent. To fill this gap, we first introduce CA-MER, a new benchmark designed to examine MLLMs under realistic emotion conflicts. It consists of three subsets: video-aligned, audio-aligned, and consistent, where only one or all modalities reflect the true emotion. However, evaluations on our CA-MER reveal that current state-of-the-art emotion MLLMs systematically over-rely on audio signal during emotion conflicts, neglecting critical cues from visual modality. To mitigate this bias, we propose MoSEAR, a parameter-efficient framework that promotes balanced modality integration. MoSEAR consists of two modules: (1)MoSE, modality-specific experts with a regularized gating mechanism that reduces modality bias in the fine-tuning heads; and (2)AR, an attention reallocation mechanism that rebalances modality contributions in frozen backbones during inference. Our framework offers two key advantages: it mitigates emotion conflicts and improves performance on consistent samples-without incurring a trade-off between audio and visual modalities. Experiments on multiple benchmarks-including MER2023, EMER, DFEW, and our CA-MER-demonstrate that MoSEAR achieves state-of-the-art performance, particularly under modality conflict conditions.

Large language models (LLMs) enable diverse forms of AI-assisted creation, yet they often struggle to bridge the preference-articulation gap: users may provide incomplete or vague intentions or lack the vocabulary to specify what they want, yielding outputs misaligned with true preferences. To address this gap and facilitate music creation in a vibe-centric environment, we introduce VibeMus, a proactive agentic system built on open-source components. The system engages in multi-turn dialogue to progressively determine the music’s emotion, genre, lyrics, and other aspects before generation. Simulated evaluations show that proactive clarification improves alignment with users’ intended nuances. Our approach is training-free, leveraging an open-source music model, an open-source agentic framework, and publicly available LLM APIs. We release our code, showcase several demos, and provide additional details at https://github.com/tuteng0915/VibeMus.

Achieving ID-preserving text-to-video (T2V) generation remains challenging despite recent advances in diffusion-based models. Existing approaches often fail to capture fine-grained facial dynamics or maintain temporal identity coherence. To address these limitations, we propose MoCA, a novel Video Diffusion Model built on a Diffusion Transformer (DiT) backbone, incorporating a Mixture of Cross-Attention mechanism inspired by the Mixture-of-Experts paradigm. Our framework improves inter-frame identity consistency by embedding MoCA layers into each DiT block, where Hierarchical Temporal Pooling captures identity features over varying timescales, and Temporal-Aware Cross-Attention Experts dynamically model spatiotemporal relationships. We further incorporate a Latent Video Perceptual Loss to enhance identity coherence and fine-grained details across video frames. To train this model, we collect CelebIPVid, a dataset of 10,000 high-resolution videos from 1,000 diverse individuals, promoting cross-ethnic generalization. Extensive experiments on CelebIPVid show that MoCA outperforms existing T2V methods by over 5% across facial similarity.

Consistent pose‐driven character animation has achieved remarkable progress in single‐character scenarios. However, extending these advances to multi‐character settings is non‐trivial, especially when position swap is involved. Beyond mere scaling, the core challenge lies in enforcing correct Identity Correspondence (IC) between characters in reference and generated frames. To address this, we introduce EverybodyDance, a systematic solution targeting IC correctness in multi-character animation. EverybodyDance is built around the **Identity Matching Graph (IMG)**, which models characters in the generated and reference frames as two node sets in a weighted complete bipartite graph. Edge weights, computed via our proposed Mask–Query Attention (MQA), quantify the affinity between each pair of characters. Our key insight is to formalize IC correctness as a graph structural metric and to optimize it during training. We also propose a series of targeted strategies tailored for multi-character animation, including identity-embedded guidance, a multi-scale matching strategy, and pre-classified sampling, which work synergistically. Finally, to evaluate IC performance, we curate the **Identity Correspondence Evaluation** benchmark, dedicated to multi‐character IC correctness. Extensive experiments demonstrate that EverybodyDance substantially outperforms state‐of‐the‐art baselines in both IC and visual fidelity.

Effective human-agent collaboration in physical environments requires understanding not only what to act upon, but also where the actionable elements are and how to interact with them. Existing approaches often operate at the object level or disjointedly handle fine-grained affordance reasoning, lacking coherent, instruction-driven grounding and reasoning. In this work, we introduce a new task: Fine-grained 3D Embodied Reasoning, which requires an agent to predict, for each referenced affordance element in a 3D scene, a structured triplet comprising its spatial location, motion type, and motion axis, based on a task instruction. To solve this task, we propose AffordBot, a novel framework that integrates Multimodal Large Language Models (MLLMs) with a tailored chain-of-thought (CoT) reasoning paradigm. To bridge the gap between 3D input and 2D-compatible MLLMs, we render surround-view images of the scene and project 3D element candidates into these views, forming a rich visual representation aligned with the scene geometry. Our CoT pipeline begins with an active perception stage, prompting the MLLM to select the most informative viewpoint based on the instruction, before proceeding with step-by-step reasoning to localize affordance elements and infer plausible interaction motions. Evaluated on the SceneFun3D dataset, AffordBot achieves state-of-the-art performance, demonstrating strong generalization and physically grounded reasoning with only 3D point cloud input and MLLMs. Our code is available at [https://github.com/hannahwxy/AffordBot](https://github.com/hannahwxy/AffordBot).

We present EgoBlind, the first egocentric VideoQA dataset collected from blind individuals to evaluate the assistive capabilities of contemporary multimodal large language models (MLLMs). EgoBlind comprises 1,392 first-person videos from the daily lives of blind and visually impaired individuals. It also features 5,311 questions directly posed or verified by the blind to reflect their in-situation needs for visual assistance. Each question has an average of 3 manually annotated reference answers to reduce subjectiveness.Using EgoBlind, we comprehensively evaluate 16 advanced MLLMs and find that all models struggle. The best performers achieve an accuracy near 60\%, which is far behind human performance of 87.4\%. To guide future advancements, we identify and summarize major limitations of existing MLLMs in egocentric visual assistance for the blind and explore heuristic solutions for improvement. With these efforts, we hope that EgoBlind will serve as a foundation for developing effective AI assistants to enhance the independence of the blind and visually impaired. Data and code are available at \url{https://github.com/doc-doc/EgoBlind}.

Recent advancements in sequential recommendation have underscored the potential of Large Language Models (LLMs) for enhancing item embeddings. However, existing approaches face three key limitations: 1) the degradation of the semantic space when high-dimensional language embeddings are mapped to lower-dimensional ID embeddings, 2) the underutilization of language embeddings, and 3) the reliance on additional trainable parameters, such as an adapter, to bridge the gap between the semantic and behavior spaces. In this paper, we introduce AlphaFuse, a simple but effective language-guided learning strategy that addresses these challenges by learning ID embeddings within the null space of language embeddings. Specifically, we decompose the semantic space of language embeddings via Singular Value Decomposition (SVD), distinguishing it into a semantic-rich row space and a semantic-sparse null space. Collaborative signals are then injected into the null space, while preserving the rich semantics of the row space. AlphaFuse prevents degradation of the semantic space, integrates the retained language embeddings into the final item embeddings, and eliminates the need for auxiliary trainable modules, enabling seamless adaptation to any sequential recommendation framework. We validate the effectiveness and flexibility of AlphaFuse through extensive experiments on three benchmark datasets, including cold-start user and long-tail settings, showcasing significant improvements in both discriminative and diffusion-based generative sequential recommenders.

Inspired by the activity-silent and persistent activity mechanisms in human visual perception biology, we design a Unified Static and Dynamic Network (UniSDNet), to learn the semantic association between the video and text/audio queries in a cross-modal environment for efficient video grounding. For static modeling, we devise a novel residual structure (ResMLP) to boost the global comprehensive interaction between the video segments and queries, achieving more effective semantic enhancement/supplement. For dynamic modeling, we effectively exploit three characteristics of the persistent activity mechanism in our network design for a better video context comprehension. Specifically, we construct a diffusely connected video clip graph on the basis of 2D sparse temporal masking to reflect the "short-term effect" relationship. We innovatively consider the temporal distance and relevance as the joint "auxiliary evidence clues" and design a multi-kernel Temporal Gaussian Filter to expand the context clue into high-dimensional space, simulating the "complex visual perception", and then conduct element level filtering convolution operations on neighbour clip nodes in message passing stage for finally generating and ranking the candidate proposals. Our UniSDNet is applicable to both Natural Language Video Grounding (NLVG) and Spoken Language Video Grounding (SLVG) tasks. Our UniSDNet achieves SOTA performance on three widely used datasets for NLVG, as well as three datasets for SLVG, e.g., reporting new records at 38.88% $R@1,[email protected]$R@1,[email protected] on ActivityNet Captions and 40.26% $R@1,[email protected]$R@1,[email protected] on TACoS. To facilitate this field, we collect two new datasets (Charades-STA Speech and TACoS Speech) for SLVG task. Meanwhile, the inference speed of our UniSDNet is 1.56× faster than the strong multi-query benchmark.

Mental health has attracted substantial attention in recent years and large language model (LLM) can be an effective technology for alleviating this problem owing to its capability in text understanding and dialogue. However, existing research in this domain often suffers from limitations, such as training on datasets lacking crucial prior knowledge and evidence, and the absence of comprehensive evaluation methods. In this article, we propose a specialized psychological LLM, named PsycoLLM, trained on a proposed high-quality psychological dataset, including single-turn QA, multiturn dialogues, and knowledge-based QA. Specifically, we construct multi-turn dialogues through a three-step pipeline comprising multiturn QA generation, evidence judgment, and dialogue refinement. We augment this process with real-world psychological case backgrounds extracted from online platforms, enhancing the relevance and applicability of the generated data. Additionally, to compare the performance of PsycoLLM with other LLMs, we develop a comprehensive psychological benchmark based on authoritative psychological counseling examinations in China, which includes assessments of professional ethics, theoretical proficiency, and case analysis. The experimental results on the benchmark illustrate the effectiveness of PsycoLLM, which demonstrates superior performance compared with other LLMs.

The emotional support conversation (ESC) system is tailored to reduce the distress of individuals who are experiencing emotional challenges. To achieve this goal, the system must fully understand the emotional state of the user and employ appropriate strategies to provide effective emotional support. However, most existing methods predominantly focus on the user’s psychological state and the causes of user’s emotional problem, neglecting key cues, e.g., topics of conversation and listener’s psychological state. This leads to an incomplete understanding of the user’s situation and ineffective emotional support. This article presents a novel method that utilizesprompt learning withmultiperspectivecues (PMPC) to generate emotional support responses. Specifically, we extract multiperspective cues from the dialogue history and the causes of user’s emotional problem. To fully utilize the different cues within the ESC system, we categorize them into two groups:semantic enhancement cuesandsemantic constraint cues. Subsequently, we construct two prompts based on different categories of cues, which are designed to thoroughly understand user’s emotional predicament and generate an engaging support response, respectively. Experimental results on the ESConv dataset demonstrate that our proposed PMPC can surpass other approaches in both automatic and human evaluation metrics, providing compelling evidence of its efficacy and potential impact in real-world applications.

Video corpus moment retrieval (VCMR) aims to retrieve a moment from a large corpus of untrimmed videos corresponding to a given language query. However, existing methods often fall short due to their reliance on simple cross-modal attention mechanisms and one-stop localization, which fail to handle the complex multimodal information and large search space effectively. To address these challenges, we propose a novel VCMR method with Query-specific Context Learning and Progressive Localization (QCLPL). First, we construct query-specific multimodal contexts that capture complementary and consistent semantics across subtitles and frames, ensuring informative and efficient context building. We further introduce a semantic contrastive loss to refine these multimodal contexts, filtering out query-irrelevant information. Additionally, we introduce a progressive localization strategy that transforms the moment localization task into a two-stage process. By classifying frames into foreground and background regions, we present a simplified binary classification problem before boundary prediction, constrained by a region-aware loss. This progressive approach leverages region priors to improve subsequent moment localization. Extensive experiments on the TVR and DiDeMo datasets demonstrate that our method significantly outperforms existing approaches, setting a new state of the art for VCMR.

This article introduces a novel extension of the multivariate variational mode decomposition (MVMD) method, termed fuzzy MVMD (FMVMD), designed to enhance alignment information extraction. In contrast to MVMD, FMVMD focuses on capturing finer alignment details by leveraging fuzzy clustering techniques. The proposed FMVMD algorithm proceeds through the following steps: First, FMVMD employs a modified clustering algorithm, termed fuzzy C-means (FCM), to categorize submodes within each channel into fuzzy clusters based on their contribution to common center frequencies. Second, a variational optimization model is formulated, extending the principles of MVMD to accommodate the fuzzy clustering approach used in FMVMD. Finally, an optimization technique called the alternating direction method of multipliers is employed to derive the optimal solution for the FMVMD model. Experimental results show that FMVMD achieves a 41% and 28% improvement in center frequency alignment performance compared to MVMD when using two and three fuzzy clusters, respectively, and a 13% improvement compared to GMVMD with the same number of clusters. Under a 25 dB SNR condition, FMVMD demonstrates a noise resistance improvement of 44% and 24% compared to MVMD with two and three fuzzy clusters, respectively, and a 37% improvement compared to GMVMD. Validation using EEG data in the forms of bipolar leads and common average reference confirms the effectiveness of FMVMD, achieving consistently favorable results.

The evolution of advanced artificial intelligence generated content approaches has heightened concerns about deepfake, due to the sophisticated forgeries and concealed appearances they produce. To this end, the pre-trained Vision Transformer (ViT) model has become a de facto choice for deepfake detection, thanks to its powerful learning capability. Despite favorable results achieved by existing ViT-based methods, they have inherent limitations that could result in suboptimal performance in scenarios with continuously evolving forgery techniques, such as overfitting to single forgery patterns or placing excessive emphasis on dominant forgery regions. In this paper, we propose CUTA, a simple yet effective deepfake detection paradigm that utilizes ViT adapters as the medium and fully exploits the spatial- and frequency-domain features of given images to overcome the limitations of existing methods. Specifically, CUTA focuses onfrequency domain maskingwithin the input space, which obscures parts of the high-frequency image to intensify the training challenge while preserving subtle forgery cues in the frequency domain to facilitate comprehensive forgery representations. Furthermore, we propose two task-customized modules within the ViT model, i.e., thetexture enhancement moduleand themulti-scale perceptron module, to seamlessly integrate local texture and rich contextual features. These two modules ensure an organic interaction between the task-specific forgery patterns and general semantic features within the pre-trained ViT framework. The experimental results on several publicly available benchmark datasets demonstrate CUTA’s superiority in performance, particularly showcasing its significant advantages in both cross-dataset and cross-manipulation scenarios.

Domain generalization (DG) aims to solve the problem of significant performance degradation when target domain data collected from the Out-Of-Distribution (O.O.D). Previous efforts try to exploit invariant features in the source domain through CNN networks. However, inspired by causal mechanisms, we find that the complex spurious-invariant information is still hidden in this view invariant features, and the impact of domain and class discrepancies on extracting invariance has not been effectively mitigated. To alleviate these issues, we propose a self-weighted multi-view mining invariance domain generalization framework (SMIDG). On the one hand, to make up for the insufficiency of traditional single-view convolutional feature extraction networks, we propose to mine features from another frequency view and use the self-adaptive adversarial masks to eliminate some spurious correlations, ensuring causal invariance in the coarse-grained generalization. However, due to inconsistencies in discriminative information between inter-domain and intra-domain samples, as well as inter-class and intra-class samples, the coarse-grained elimination of spurious associations does not fully resolve this issue. On the other hand, we also consider the fine-grained generalization from two aspects. Firstly, to better tackle the domain discrepancies, we propose a novel progressive contrastive learning strategy that learns the underlying specific features of samples while gradually mitigating domain discrepancies, thereby ensuring domain invariance in fine-grained generalization. Secondly, due to the issue of feature inconsistency, we adopt a self-adaptive hard sample mining method with information gain to ensure that the model pays more attention on hard disentangled samples, thus maintaining feature invariance. Extensive experiments on five benchmark datasets demonstrate that our method outperforms state-of-the-art approaches. Our code is available at https://github.com/bihhm/SMIDG.

Deep neural networks (DNNs) have achieved remarkable success in various computer vision tasks, e.g., image classification. However, most of the existing models depend heavily on annotated data, where label noise is inevitable. Training with such noisy data negatively impacts the generalization performance of DNNs. To this end, recent advances in learning with noisy labels (LNL) adopt the sample selection strategy that identifies clean samples from the noisy dataset to update DNNs, using semi-supervised learning where rejected samples are treated as unlabeled data. However, existing LNL methods often overlook the varying fitting difficulties of different classes, resulting in suboptimal sample selection and confirmation bias, and consequently, the errors accumulate during semi-supervised training. In this article, we propose a novel method, TNCollab, which aims at alleviating confirmation bias in both sample selection and semi-supervised training stages via two-network collaboration. Specifically, we introduce a class-adaptive threshold for sample selection to address the varying fitting difficulties across different classes. Additionally, we construct a hard set consisting of samples where the two networks disagree and introduce a noise-robust loss to extract potentially useful information while maintaining robustness against label noise. Furthermore, we propose a dual consistency loss to ensure consistent predictions between the networks across different augmented views of the same sample, facilitating mutual learning. Extensive experiments demonstrate that TNCollab achieves state-of-the-art performance on image classification and facial expression recognition tasks, particularly on CIFAR-10, CIFAR-100, WebVision, Clothing1M, Tiny-ImageNet, and RAF-DB datasets, showing improved visual understanding and generalization capabilities. Our codes are available at https://github.com/Delete12137/TNCollab .

Knowledge tracing has been widely used in online learning systems to guide the students’ future learning. However, most existing KT models primarily focus on extracting abundant information from the question sets and explore the relationships between them, but ignore the personalized student behavioral information in the learning process. This will limit the model’s ability to accurately capture the personalized knowledge states of students and reasonably predict their performances. To alleviate this limitation, we explicitly models the personalized learning process by incorporating the emotions, a representative personalized behavior in the learning process, into KT framework. Specifically, we present a novel Dual-State Personalized Knowledge Tracing with Emotional Incorporation model to achieve this goal: First, we incorporate emotional information into the modeling process of knowledge state, resulting in the Knowledge State Boosting Module. Second, we design an Emotional State Tracing Module to monitor students’ personalized emotional states, and propose an emotion prediction method based on personalized emotional states. Finally, we apply the predicted emotions to enhance students’ response prediction. Furthermore, to extend the generalization capability of our model across different datasets, we design a transferred version of DEKT, named Transfer Learning-based Self-loop model (T-DEKT). Extensive experiments show our method achieves the state-of-the-art performance.

Repetitive Action Counting (RAC) aims to count the number of repetitive actions occurring in videos. In the real world, repetitive actions have great diversity and bring numerous challenges (e.g., viewpoint changes, non-uniform periods, and action interruptions). Existing methods based on the temporal self-similarity matrix (TSSM) for RAC are trapped in the bottleneck of insufficient capturing action periods when applied to complicated daily videos. To tackle this issue, we propose a novel method named Hybrid Temporal Relation Modeling Network (HTRM-Net) to build diverse TSSM for RAC. The HTRM-Net mainly consists of three key components: bi-modal temporal self-similarity matrix modeling, random matrix dropping, and local temporal context modeling. Specifically, we construct temporal self-similarity matrices by bi-modal (self-attention and dual-softmax) operations, yielding diverse matrix representations from the combination of row-wise and column-wise correlations. To further enhance matrix representations, we propose incorporating a random matrix dropping module to guide channel-wise learning of the matrix explicitly. After that, we inject the local temporal context of video frames and the learned matrix into temporal correlation modeling, which can make the model robust enough to cope with error-prone situations, such as action interruption. Finally, a multi-scale matrix fusion module is designed to aggregate temporal correlations adaptively in multi-scale matrices. Extensive experiments across intra- and cross-datasets demonstrate that the proposed method not only outperforms current state-of-the-art methods and but also exhibits robust capabilities in accurately counting repetitive actions in unseen action categories. Notably, our method surpasses the classical TransRAC method by 20.04% in MAE and 22.76% in OBO.

Despite significant strides in the field of 3D scene editing, current methods encounter substantial challenge, particularly in preserving 3D consistency during the multi-view editing process. To tackle this challenge, we propose a progressive 3D editing strategy that ensures multi-view consistency via a Trajectory-Anchored Scheme (TAS) with a dual-branch editing mechanism. Specifically, TAS facilitates a tightly coupled iterative process between 2D view editing and 3D updating, preventing error accumulation yielded from the text-to-image process. Additionally, we explore the connection between optimization-based methods and reconstruction-based methods, offering a unified perspective for selecting superior design choices, supporting the rationale behind the designed TAS. We further present a tuning-free View-Consistent Attention Control (VCAC) module that leverages cross-view semantic and geometric reference from the source branch to yield aligned views from the target branch during the editing of 2D views. To validate the effectiveness of our method, we analyze 2D examples to demonstrate the improved consistency with the VCAC module. Extensive quantitative and qualitative results in text-guided 3D scene editing clearly indicate that our method can achieve superior editing quality compared with state-of-the-art 3D scene editing methods. Our project site is athttps://fkcptlst.github.io/TrAME/

Partially relevant video retrieval (PRVR) is a practical yet challenging task in text-to-video retrieval, where videos are untrimmed and contain much background content. The pursuit here is of both effective and efficient solutions to capture the partial correspondence between text queries and untrimmed videos. Existing PRVR methods, which typically focus on modeling multi-scale clip representations, however, suffer from content independence and information redundancy, impairing retrieval performance. To overcome these limitations, we propose a simple yet effective approach with active moment discovering (AMDNet). We are committed to discovering video moments that are semantically consistent with their queries. By using learnable span anchors to capture distinct moments and applying masked multi-moment attention to emphasize salient moments while suppressing redundant backgrounds, we achieve more compact and informative video representations. To further enhance moment modeling, we introduce a moment diversity loss to encourage different moments of distinct regions and a moment relevance loss to promote semantically query-relevant moments, which cooperate with a partially relevant retrieval loss for end-to-end optimization. Extensive experiments on two large-scale video datasets (i.e., TVR and ActivityNet Captions) demonstrate the superiority and efficiency of our AMDNet. In particular, AMDNet is about 15.5 times smaller (#parameters) while 6.0 points higher (SumR) than the up-to-date method GMMFormer on TVR.

While the diffusion transformer (DiT) has become a focal point of interest in recent years, its application in low-light image enhancement remains a blank area for exploration. Current methods recover the details from low-light images while inevitably amplifying the noise in images, resulting in poor visual quality. In this paper, we firstly introduce DiT into the low-light enhancement task and design a novel Structure-guided Diffusion Transformer based Low-light image enhancement (SDTL) framework. We compress the feature through wavelet transform to improve the inference efficiency of the model and capture the multi-directional frequency band. Then we propose a Structure Enhancement Module (SEM) that uses structural prior to enhance the texture and leverages an adaptive fusion strategy to achieve more accurate enhancement effect. In Addition, we propose a Structure-guided Attention Block (SAB) to pay more attention to texture-riched tokens and avoid interference from noisy areas in noise prediction. Extensive qualitative and quantitative experiments demonstrate that our method achieves SOTA performance on several popular datasets, validating the effectiveness of SDTL in improving image quality and the potential of DiT in low-light enhancement tasks.

Referring Image Segmentation (RIS) has been extensively studied over the past decade, leading to the development of advanced algorithms. However, there has been a lack of research investigating how existing algorithms should be benchmarked with complex language queries, which include more informative descriptions of surrounding objects and backgrounds (e.g., the black car vs. the black car is parking on the road and beside the bus ). Given the significant improvement in the semantic understanding capability of large pre-trained models, it is crucial to take a step further in RIS by incorporating complex language that resembles real-world applications. To close this gap, building upon the existing RefCOCO and Visual Genome datasets, we propose a new RIS benchmark with complex queries, namely RIS-CQ . The RIS-CQ dataset is of high quality and large scale, which challenges the existing RIS with enriched, specific, and informative queries, and enables a more realistic scenario of RIS research. Besides, we present a niche targeting method to better task the RIS-CQ, called Dual-Modality Graph Alignment ( DuMoGa ) model, which outperforms a series of RIS methods. To provide a valuable foundation for future advancements in the field of RIS with complex queries, we release the datasets, pre-processing and synthetic scripts, and the algorithm implementations at https://github.com/lili0415/DuMoGa .

Hyperspectral imaging is a valuable technique for accurately classifying materials because of the abundance of spectral information and high resolution it provides. However, the characteristics of Hyperspectral Imaging, such as high-dimensional features and information redundancy, pose significant challenges to data processing. Traditional dimensionality reduction methods often have information loss, high computational complexity, and easy to ignore the strong correlation between HSI bands when dealing with the HSI data. Although other methods can achieve satisfactory classification performance, they do not consider the dimensionality reduction of HSI, and they focus on the model performance, which limits further improvement in classification performance. This article proposes a transformer-based framework called “SpectrumRecombineFormer” (SRF), which is composed of two key modules, namely “Spatial–Spectral Recombination” (SSRC) and “Cross-Layer Fusion” (CF). The SSRC is capable of utilizing both adjacent and non-adjacent spectrums to generate the spatial-sequential perceptive representations, which alleviate the effect of the strong correlation between HSI bands. The CF can avoid the loss of information during the feed-forward procedure among layers. Extensive experiments on five existing datasets (widely adopted Indian Pines, Houston2013, Pavia University, Salinas, and KSC) demonstrate the capability of our proposed method to address the above-mentioned challenges. Both quantitative and qualitative experimental ablation studies, including visualization results, reveal that the proposed SRF method can successfully and efficiently classify HSIs and surpass the other state-of-the-art methods. For access to the source code, please visit https://github.com/kangpeilun/SRF-HSI-Classification-master .

The remarkable progress in cross-modal retrieval relies on accurately annotated multimedia datasets. In practice, most existing datasets used for training cross-modal retrieval models are automatically collected from the Internet to reduce data collection costs. However, it inevitably contains mismatched pairs, i.e., noisy correspondences, thus degrading the model performance. Recent advances utilize the predicted similarity distribution of individual samples for noise validation and correction, which easily faces two challenging dilemmas: (1) confirmation bias and (2) unstable performance with increasing noise. In light of the above, we propose a generalized Bias Mitigation and Representation Optimization (BMRO) framework. Specifically, we propose a Bias Estimator (BE) to estimate the unbiased confidence factor of a sample by contrasting it against its nearest neighbors. Unbiased confidence factor can precisely adjust sample contribution and enhance accurate sample division. This facilitates the Adaptive Representation Optimizer (ARO) in providing tailored optimization strategies for clean and noisy samples. ARO performs contrastive learning between clean samples and generated hard samples, thus promoting the generalizability and robustness of the representation. Besides, it utilizes complementary learning to reduce incorrect guidance from noisy samples. Extensive experiments on five visual-text benchmarks verify that our BMRO can significantly improve the matching accuracy and performance stability against noisy correspondences.

Our life is populated with articulated objects. Current category-level articulation estimation works largely focus on predicting part-level 6D poses on static point cloud observations. In this paper, we tackle the problem of category-level online robust and real-time 6D pose tracking of articulated objects, where we propose KPA-Tracker, a novel 3D KeyPoint based Articulated object pose Tracker. Given an RGB-D image or a partial point cloud at the current frame as well as the estimated per-part 6D poses from the last frame, our KPA-Tracker can effectively update the poses with learned 3D keypoints between the adjacent frames. Specifically, we first canonicalize the input point cloud and formulate the pose tracking as an inter-frame pose increment estimation task. To learn consistent and separate 3D keypoints for every rigid part, we build KPA-Gen that outputs the high-quality ordered 3D keypoints in an unsupervised manner. During pose tracking on the whole video, we further propose a keypoint-based articulation tracking algorithm that mines keyframes as reference for accurate pose updating. We provide extensive experiments on validating our KPA-Tracker on various datasets ranging from synthetic point cloud observation to real-world scenarios, which demonstrates the superior performance and robustness of the KPA-Tracker. We believe that our work has the potential to be applied in many fields including robotics, embodied intelligence and augmented reality. All the datasets and codes are available at https://github.com/hhhhhar/KPA-Tracker.

Text-based Person Retrieval (TPR) aims to retrieve relevant images of specific pedestrians based on the given textual query. The mainstream approaches primarily leverage pretrained deep neural networks to learn the mapping of visual and textual modalities into a common latent space for cross-modality matching. Despite their remarkable achievements, existing efforts mainly focus on learning the statistical cross-modality correlation found in training data, other than the intrinsic causal correlation. As a result, they often struggle to retrieve accurately in the face of environmental changes such as illumination, pose, and occlusion, or when encountering images with similar attributes. In this regard, we pioneer the observation of TPR from a causal view. Specifically, we assume that each image is composed of a mixture of causal factors (which are semantically consistent with text descriptions) and non-causal factors (retrieval-irrelevant, e.g., background), and only the former can lead to reliable retrieval judgments. Our goal is to extract text-critical robust visual representation (i.e., causal factors) and establish domain invariant cross-modality correlations for accurate and reliable retrieval. However, causal/non-causal factors are unobserved, so we emphasize that ideal causal factors that can simulate causal scenes should satisfy two basic principles:1） Independence: being independent of non-causal factors, and 2）Sufficiency: being causally sufficient for TPR across different environments. Building on that, we propose an Invariant Representation Learning method for TPR (IRLT), that enforces the visual representations to satisfy the two aforementioned critical properties. Extensive experiments on three datasets clearly demonstrate the advantages of IRLT over leading baselines in terms of accuracy and generalization.

Cognitive Diagnosis Modeling aims to infer students' proficiency level on knowledge concepts from their response logs. Existing methods typically model students’ response processes as the interaction between students and exercises or concepts based on hand-crafted or deeply-learned interaction functions. Despite their promising achievements, they fail to consider the relationship between students' cognitive states and affective states in learning, e.g., the feelings of frustration, boredom, or confusion with the learning content, which is insufficient for comprehensive cognitive diagnosis in intelligent education. To fill the research gap, we propose a novel Affect-aware Cognitive Diagnosis (ACD) model which can effectively diagnose the knowledge proficiency levels of students by taking into consideration the affective factors. Specifically, we first design a student affect perception module under the assumption that the affective state is jointly influenced by the student's affect trait and the difficulty of the exercise. Then, our inferred affective distribution is further used to estimate the student's subjective factors, i.e., guessing and slipping, respectively. Finally, we integrate the estimated guessing and slipping parameters with the basic neural cognitive diagnosis framework based on the DINA model, which facilitates the modeling of complex exercising interactions in a more accurate and interpretable fashion. Besides, we also extend our affect perception module in an unsupervised learning setting based on contrastive learning, thus significantly improving the compatibility of our ACD. To the best of our knowledge, we are the first to unify the cognition modeling and affect modeling into the same framework for student cognitive diagnosis. Extensive experiments on real-world datasets clearly demonstrate the effectiveness of our ACD. Our code is available at https://github.com/zeng-zhen/ACD.

Long-term and accurate forecasting is the long-standing pursuit of the human motion prediction task. Existing methods typically suffer from dramatic degradation in prediction accuracy with increasing prediction horizon. It comes down to two reasons: 1) Insufficient numerical stability caused by unforeseen high noise and complex feature relationships in the data, and 2) Inadequate modeling stability caused by unreasonable step sizes and undesirable parameter updates in the prediction. In this paper, we design a novel and sym-plectic integral-inspired framework named symplectic integral neural network (SINN), which engages symplectic tra-jectories to optimize the pose representation and employs a stable symplectic operator to alternately model the dynamic context. Specifically, we design a Symplectic Repre-sentation Encoder that performs on enhanced human pose representation to obtain trajectories on the symplectic manifold, ensuring numerical stability based on Hamiltonian mechanics and symplectic spatial splitting algorithm. We further present the Symplectic Temporal Aggregation mod-ule, which splits the long-term prediction into multiple ac-curate short-term predictions generated by a symplectic operator to secure modeling stability. Moreover, our approach is model-agnostic and can be efficiently integrated with different physical dynamics models. The experimental results demonstrate that our method achieves the new state-of-the-art, outperforming existing methods by 20.1% on Human3.6M, 16.7% on CUM Mocap, and 10.2% on 3DPW.

One-shot Federated Learning (OFL) has become a promising learning paradigm, enabling the training of a global server model via a single communication round. In OFL, the server model is aggregated by distilling knowledge from all client models (the ensemble), which are also responsible for synthesizing samples for distillation. In this regard, advanced works show that the performance of the server model is intrinsically related to the quality of the synthesized data and the ensemble model. To promote OFL, we introduce a novel framework, Co-Boosting, in which synthesized data and the ensemble model mutually enhance each other progressively. Specifically, Co-Boosting leverages the current ensemble model to synthesize higher-quality samples in an adversarial attack manner. These hard samples are then employed to promote the quality of the ensemble model by adjusting the ensembling weights for each client model. Consequently, Co-Boosting periodically achieves high-quality data and ensemble models. Extensive experiments demonstrate that Co-Boosting can substantially outperform existing baselines under various settings. Moreover, Co-Boosting eliminates the need for adjustments to the client's local training, requires no additional data or model transmission, and allows client models to have heterogeneous architectures.

Prompt learning represents a promising method for adapting pre-trained vision-language models (VLMs) to various downstream tasks by learning a set of text embeddings. One challenge inherent to these methods is the poor generalization performance due to the invalidity of the learned text embeddings for unseen tasks. A straightforward approach to bridge this gap is to freeze the text embeddings in prompts, which results in a lack of capacity to adapt VLMs for downstream tasks. To address this dilemma, we propose a paradigm called EnPrompt with a novel External Layer (EnLa). Specifically, we propose a textual external layer and learnable visual embeddings for adapting VLMs to downstream tasks. The learnable external layer is built upon valid embeddings of pre-trained CLIP. This design considers the balance of learning capabilities between the two branches. To align the textual and visual features, we propose a novel two-pronged approach: i) we introduce the optimal transport as the discrepancy metric to align the vision and text modalities, and ii) we introduce a novel strengthening feature to enhance the interaction between these two modalities. Four representative experiments (i.e., base-to-novel generalization, few-shot learning, cross-dataset generalization, domain shifts generalization) across 15 datasets demonstrate that our method outperforms the existing prompt learning method.

Deep Cross-Modal Hashing (CMH) has become one of the most popular solutions for cross-modal retrieval. Existing methods need to first collect data and then be trained with these accumulated data. However, in real world, data may be generated and possessed by different owners. Considering the concerns about privacy, data may not be shared or transmitted, leading to the failure of sufficient training of CMH. To solve the problem, we propose a new framework called Federated Cross-modal Hashing with Adaptive Feature Enhancement (FedCAFE). FedCAFE is a federated method which could use distributed data to train existing CMH methods under the privacy protection. To overcome the data heterogeneity challenge of distributed data and improve the generalization ability of global model, FedCAFE is endowed with a novel adaptive feature enhancement module and a new weighted aggregation strategy. Besides, it could fully utilize the rich global information carried in the global model to constrain the model during the local training process. We have conducted extensive experiments on four widely-used datasets in CMH domain with both IID and non-IID settings. The reported results demonstrate that the proposed FedCAFE achieves better performance than several state-of-the-art baselines.

Video Moment Retrieval (MR) tasks involve predicting the moment described by a given natural language or spoken language query in an untrimmed video. In this paper, we propose a novel Maskable Retentive Network (MRNet) to address two key challenges in MR tasks: cross-modal guidance and video sequence modeling. Our approach introduces a new retention mechanism into the multimodal Transformer architecture, incorporating modality-specific attention modes. Specifically, we employ the Unlimited Attention for language-related attention regions to maximize cross-modal mutual guidance. Then, we introduce the Maskable Retention for video-only attention region to enhance video sequence modeling, that is, recognizing two crucial characteristics of video sequences: 1) bidirectional, decaying, and non-linear temporal associations between video clips, and 2) sparse associations of key information semantically related to the query. We propose a bidirectional decay retention mask to explicitly model temporal-distant context dependencies of video sequences, along with a learnable sparse retention mask to adaptively capture strong associations relevant to the target event. Extensive experiments conducted on five popular benchmarks ActivityNet Captions, TACoS, Charades-STA, ActivityNet Speech, and QVHighlights for MR tasks demonstrate the significant improvements achieved by our method over existing approaches. Code is available at https://github.com/xian-sh/MRNet.

Unpaired point cloud completion involves filling in missing parts of a point cloud without requiring partial-complete correspondence. Meanwhile, since point cloud completion is an ill-posed problem, there are multiple ways to generate the missing parts. Existing unpaired completion methods usually leverage generative adversarial training by transforming partial shape encoding into a complete one in the low-dimensional latent feature space. However, "mode collapse" often occurs, where only a subset of the shapes is represented in the low-dimensional space, reducing the diversity of the generated shapes. In this paper, we propose a novel unpaired multimodal shape completion approach that directly operates on point coordinate space. We achieve unpaired completion via a single diffusion model trained on complete data by "hijacking" the generative process. We further augment the diffusion model by introducing two guidance mechanisms to facilitate mapping the partial point cloud to the complete one while preserving its original structure. We conduct extensive evaluations of our approach, which show that our method generates shapes that are more diverse and better preserve the original structures compared to alternative methods.

Point cloud plays a significant role in recent learning-based vision tasks, which contain additional information about the physical space compared to 2D images. However, such a 3D data format also results in more expensive training costs to train a sophisticated network with large 3D datasets. Previous methods for point cloud compression focus on compacting the representation of each point cloud for better storage and transmission but ignore the improvements in training efficiency. In this paper, we introduce a new open problem in the point cloud field, named point cloud condensation : Can we condense a large point cloud dataset into a much smaller synthetic dataset while preserving the important information of the original large dataset? In other words, we explore the possibility of training a network on a smaller dataset of informative point clouds extracted from the original large dataset but maintaining similar network classification performance. Training on this small synthetic dataset could largely improve the training efficiency. To achieve this goal, we propose a two-stage approach to generate the synthetic dataset. We first introduce a nearest-feature-mean based strategy to initialize the synthetic dataset, and then formulate our goal as a parameter-matching problem, which we solve by introducing a gradient-matching strategy to iteratively refine the synthetic dataset. We conduct extensive experiments on various synthetic and real-scanned 3D object classification benchmarks, showing that our synthetic dataset could achieve almost the same performance with only 5% point clouds of ScanObjectNN dataset compared to training with the full dataset. Codes are available at https://github.com/XLechter/PointCondensation.

As a widely explored multi-modal task, Temporal Sentence Grounding in videos (TSG) endeavors to retrieve a specific video segment matched with a given query text from a video. The traditional paradigm for TSG generally assumes that relevant segments always exist within a given video. However, this assumption is restrictive and unrealistic in real-world applications where the existence of a query-related segment is uncertain, easily resulting in erroneous grounding. Motivated by the research gap and practical application, this paper introduces a new task, named Temporal Sentence Grounding with Relevance Feedback (TSG-RF) in videos, which accommodates the possibility that a video may or may not include a segment related to the query. This task entails localizing precise video segments that semantically align with the query text when such content is present, while delivering definitive feedback on the non-existence of related segments when absent. Moreover, we propose a novel Relation-aware Temporal Sentence Grounding (RaTSG) network for addressing this challenging task. This network first reformulates the TSG-RF task as a foreground-background detection problem by investigating whether the query-related semantics exist in both frame and video levels. Then, a multi-granularity relevance discriminator is exploited to produce precise video-query relevance feedback and a relation-aware segment grounding module is employed to selectively conduct the grounding process, dynamically adapting to the presence or absence of query-related segments in videos. To validate our RaTSG network, we reconstruct two popular TSG datasets, establishing a rigorous benchmark for TSG-RF. Experimental results demonstrate the effectiveness of our proposed RaTSG for the TSG-RF task. Our source code is available at https://github.com/HuiGuanLab/RaTSG.

Due to factors such as motion blur, video out-of-focus, and occlusion, multi-frame human pose estimation is a challenging task. Exploiting temporal consistency between consecutive frames is an efficient approach for addressing this issue. Currently, most methods explore temporal consistency through refinements of the final heatmaps. The heatmaps contain the semantics information of key points, and can improve the detection quality to a certain extent. However, they are generated by features, and feature-level refinements are rarely considered. In this paper, we propose a human pose estimation framework with refinements at the feature and semantics levels. We align auxiliary features with the features of the current frame to reduce the loss caused by different feature distributions. An attention mechanism is then used to fuse auxiliary features with current features. In terms of semantics, we use the difference information between adjacent heatmaps as auxiliary features to refine the current heatmaps. The method is validated on the large-scale benchmark datasets PoseTrack2017 and PoseTrack2018, and the results demonstrate the effectiveness of our method.

Remote photoplethysmography (rPPG)-based physiological measurement is an emerging yet crucial vision task, whose challenge lies in exploring accurate rPPG prediction from facial videos accompanied by noises of illumination variations, facial occlusions, head movements, etc., in a noncontact manner. Existing mainstream convolutional neural network (CNN)-based models make efforts to detect physiological signals by capturing subtle color changes in facial regions of interest (ROI) caused by heartbeats. However, such models are constrained by the limited local spatial or temporal receptive fields in the neural units. Unlike them, a native transformer-based framework called dual-path TokenLearner (dual-TL) is proposed in this article, which utilizes the concept of learnable tokens to integrate both spatial and temporal informative contexts from the global perspective of the video. Specifically, the proposed dual-TL uses a spatial TokenLearner (S-TL) to explore associations in different facial ROIs, which promises the rPPG prediction far away from noisy ROI disturbances. Complementarily, a temporal TokenLearner (T-TL) is designed to infer the quasi-periodic pattern of heartbeats, which eliminates temporal disturbances such as head movements. The two TokenLearners, S-TL and T-TL, are executed in a dual-path mode. This enables the model to reduce noise disturbances for final rPPG signal prediction. Extensive experiments on four physiological measurement benchmark datasets are conducted. The dual-TL achieves state-of-the-art performances in both intra and cross-dataset testings, demonstrating its immense potential as a basic backbone for rPPG measurement.

Nuclear norm maximization has shown the power to enhance the transferability of unsupervised domain adaptation model (UDA) in an empirical scheme. In this paper, we identify a new property termedequity, which indicates the balance degree of predicted classes, to demystify the efficacy of nuclear norm maximization for UDA theoretically. With this in mind, we offer a new discriminability-and-equity maximization paradigm built on squares loss, such that predictions are equalized explicitly. To verify its feasibility and flexibility, two new losses termed Class Weighted Squares Maximization (CWSM) and Normalized Squares Maximization (NSM), are proposed to maximize both predictive discriminability and equity, from the class level and the sample level, respectively. Importantly, we theoretically relate these two novel losses (i.e., CWSM and NSM) to the equity maximization under mild conditions, and empirically suggest the importance of the predictive equity in UDA. Moreover, it is very efficient to realize the equity constraints in both losses. Experiments of cross-domain image classification on three popular benchmark datasets show that both CWSM and NSM contribute to outperforming the corresponding counterparts.

Effectively summarizing and re-expressing video content by natural languages in a more human-like fashion is one of the key topics in the field of multimedia content understanding. Despite good progress made in recent years, existing efforts usually overlooked the emotions in user-generated videos, thus making the generated sentence a bit boring and soulless. To fill the research gap, this paper presents a novel emotional video captioning framework in which we design a Vision-based Emotion Interpretation Network to effectively capture the emotions conveyed in videos and describe the visual content in both factual and emotional languages. Specifically, we first model the emotion distribution over an open psychological vocabulary to predict the emotional state of videos. Then, guided by the discovered emotional state, we incorporate visual context, textual context, and visual-textual relevance into an aggregated multimodal contextual vector to enhance video captioning. Furthermore, we optimize the network in a new emotion-fact coordinated way that involves two losses- Emotional Indication Loss and Factual Contrastive Loss, which penalize the error of emotion prediction and visual-textual factual relevance, respectively. In other words, we innovatively introduce emotional representation learning into an end-to-end video captioning network. Extensive experiments on public benchmark datasets, EmVidCap and EmVidCap-S, demonstrate that our method can significantly outperform the state-of-the-art methods by a large margin. Quantitative ablation studies and qualitative analyses clearly show that our method is able to effectively capture the emotions in videos and thus generate emotional language sentences to interpret the video content.

Vision-and-language pre-training models have achieved impressive performance for image captioning. But most of them are trained with millions of paired image-text data and require huge memory and computing overhead. To alleviate this, we try to stand on the shoulders of large-scale pre-trained language models (PLM) and pre-trained vision models (PVM) and efficiently connect them for image captioning. There are two major challenges: one is that language and vision modalities have different semantic granularity (e.g., a noun may cover many pixels), and the other is that the semantic gap still exists between the pre-trained language and vision models. To this end, we design a lightweight and efficient connector to glue PVM and PLM, which holds a criterion of selection-then-transformation . Specifically, in the selection phase, we treat each image as a set of patches instead of pixels. We select salient image patches and cluster them into visual regions to align with text. Then, to effectively reduce the semantic gap, we propose to map the selected image patches into text space through spatial and channel transformations. With training on image captioning datasets, the connector learns to bridge the semantic granularity and semantic gap via backpropagation, preparing for the PLM to generate descriptions. Experimental results on the MSCOCO and Flickr30k datasets demonstrate that our method yields comparable performance to existing works. By solely training the small connector, we achieve a CIDEr performance of 132.2% on the MSCOCO Karpathy test split. Moreover, our findings reveal that fine-tuning the PLM can further enhance performance potential, resulting in a CIDEr score of 140.6%. Code and models are available at https://github.com/YuanEZhou/PrefixCap .

Cross-lingual cross-modal retrieval aims at leveraging human-labeled annotations in a source language to construct cross-modal retrieval models for a new target language, due to the lack of manually-annotated dataset in low-resource languages (target languages). Contrary to the growing developments in the field of monolingual cross-modal retrieval, there has been less research focusing on cross-modal retrieval in the cross-lingual scenario. A straightforward method to obtain target-language labeled data is translating source-language datasets utilizing Machine Translations (MT). However, as MT is not perfect, it tends to introduce noise during translation, rendering textual embeddings corrupted and thereby compromising the retrieval performance. To alleviate this, we propose Noise-Robust Fine-tuning (NRF) which tries to extract clean textual information from a possibly noisy target-language input with the guidance of its source-language counterpart. Besides, contrastive learning involving different modalities are performed to strengthen the noise-robustness of our model. Different from traditional cross-modal retrieval methods which only employ image/video-text paired data for fine-tuning, in NRF, selected parallel data plays a key role in improving the noise-filtering ability of our model. Extensive experiments are conducted on three video-text and image-text retrieval benchmarks across different target languages, and the results demonstrate that our method significantly improves the overall performance without using any image/video-text paired data on target languages.

Recommender systems suffer from confounding biases when there exist confounders affecting both item features and user feedback (e.g.like or not). Existing causal recommendation methods typically assume confounders are fully observed and measured, forgoing the possible existence of hidden confounders in real applications. For instance, product quality is a confounder since it affects both item prices and user ratings, but is hidden for the third-party e-commerce platform due to the difficulty of large-scale quality inspection; ignoring it could result in the bias effect of over-recommending high-price items. This work analyzes and addresses the problem from a causal perspective. The key lies in modeling the causal effect of item features on a user's feedback. To mitigate hidden confounding effects, it is compulsory but challenging to estimate the causal effect without measuring the confounder. Towards this goal, we propose a Hidden Confounder Removal (HCR) framework that leverages front-door adjustment to decompose the causal effect into two partial effects, according to the mediators between item features and user feedback. The partial effects are independent from the hidden confounder and identifiable. During training, HCR performs multi-task learning to infer the partial effects from historical interactions. We instantiate HCR for two scenarios and conduct experiments on three real-world datasets. Empirical results show that the HCR framework provides more accurate recommendations, especially for less-active users. We will release the code once accepted.

Cross-modal video retrieval aims to retrieve semantically relevant videos when given a textual query, and is one of the fundamental multimedia tasks. Most top-performing methods primarily leverage Vision Transformer (ViT) to extract video features (Lei et al., 2021}, (Bain et al., 2021), (Wang et al., 2022). However, they suffer from the high computational complexity of ViT, especially when encoding long videos. A common and simple solution is to uniformly sample a small number (e.g., 4 or 8) of frames from the target video (instead of using the whole video) as ViT inputs. The number of frames has a strong influence on the performance of ViT, e.g., using 8 frames yields better performance than using 4 frames but requires more computational resources, resulting in a trade-off. To get free from this trade-off, this paper introduces an automatic video compression method based on a bilevel optimization program (BOP) consisting of both model-level (i.e., base-level) and frame-level (i.e., meta-level) optimizations. The model-level optimization process learns a cross-modal video retrieval model whose input includes the “compressed frames” learned by frame-level optimization. In turn, frame-level optimization is achieved through gradient descent using the meta loss of the video retrieval model computed on the whole video. We call this BOP method (as well as the “compressed frames”) the Meta-Optimized Frames (MOF) approach. By incorporating MOF, the video retrieval model is able to utilize the information of whole videos (for training) while taking only a small number of input frames in its actual implementation. The convergence of MOF is guaranteed by meta gradient descent algorithms. For evaluation purposes, we conduct extensive cross-modal video retrieval experiments on three large-scale benchmarks: MSR-VTT, MSVD, and DiDeMo. Our results show that MOF is a generic and efficient method that boost multiple baseline methods, and can achieve a new state-of-the-art performance.

Monocular 3D human pose estimation is an ill-posed problem in computer vision due to its depth ambiguity. Most existing works supplement the depth information by extracting temporal pose features from video frames, and they have made notable progress. However, these approaches divide a long sequence of video frames into multiple short sequences for separate processing, which leads to the loss of complementary information between sequences. Furthermore, the short-term temporal correlation among frames in a sequence is often not fully exploited. To model temporal dependencies efficiently, we propose the frame-padded multiscale transformer approach, which includes a frame-padded video sequence preprocessing step and a multiscale temporal transformer backbone. Our approach addresses the omission of the temporal features of edge frames in existing approaches by padding video frames in the shallow layer. In addition, we extract the temporal information of 3D human poses using a multiscale transformer to enhance the short-term correlation of human pose skeleton keypoints. Extensive experiments validate the effectiveness of our approach on two popular datasets: Human3.6M and MPI-INF-3DHP. The results show that our approach achieves state-of-the-art performance.

Deep image prior (DIP) is an emerging technology that indicates that the structure of an untrained network can serve as an excellent prior for image restoration. It bridges the gap between training-based and training-free methods and exhibits considerable potential in image compressed sensing (CS) reconstruction. In this article, we extend DIP and propose a novel Low-Rank Regularization Video Compressed Sensing Network for CS video reconstruction (dubbed LRR-VCSNet). We explore the application of a low-rank latent tensor with an untrained network for global low-rank regularization on video reconstruction, and the interframe low-rank approximation for framewise nonlocal low-rank regularization in the data space is also exploited. In addition, we design the structure of the untrained network based on the encoder-decoder architecture to improve the performance. Extensive experiments on six standard CIF video sequences show that LLR-VCSNet significantly outperforms traditional video CS methods and achieves competitive results when compared with the state-of-the-art training-based video CS method.

Federated learning (FL) is a promising framework for privacy-preserving and distributed training with decentralized clients. However, there exists a large divergence between the collected local updates and the expected global update, which is known as the client drift and mainly caused by heterogeneous data distribution among clients, multiple local training steps, and partial client participation training. Most existing works tackle this challenge based on the empirical risk minimization (ERM) rule, while less attention has been paid to the relationship between the global loss landscape and the generalization ability. In this work, we propose FedGAMMA, a novel FL algorithm with Global sharpness-Aware MiniMizAtion to seek a global flat landscape with high performance. Specifically, in contrast to FedSAM which only seeks the local flatness and still suffers from performance degradation when facing the client-drift issue, we adopt a local varieties control technique to better align each client's local updates to alleviate the client drift and make each client heading toward the global flatness together. Finally, extensive experiments demonstrate that FedGAMMA can substantially outperform several existing FL baselines on various datasets, and it can well address the client-drift issue and simultaneously seek a smoother and flatter global landscape.

Gaze following aims to predict where a person is looking in a scene. Existing methods tend to prioritize traditional 2D RGB visual cues or require burdensome prior knowledge and extra expensive datasets annotated in 3D coordinate systems to train specialized modules to enhance scene modeling. In this work, we introduce a novel framework deployed on a simple ResNet backbone, which exclusively uses image and depth maps to mimic human visual preferences and realize 3D-like depth perception. We first leverage depth maps to formulate spatial-based proximity information regarding the objects with the target person. This process sharpens the focus of the gaze cone on the specific region of interest pertaining to the target while diminishing the impact of surrounding distractions. To capture the diverse dependence of scene context on the saliency gaze cone, we then introduce a learnable grid-level regularized attention that anticipates coarse-grained regions of interest, thereby refining the mapping of the saliency feature to pixel-level heatmaps. This allows our model to better account for individual differences when predicting others’ gaze locations. Finally, we employ the KL-divergence loss to super the grid-level regularized attention, which combines the gaze direction, heatmap regression, and in/out classification losses, providing comprehensive supervision for model optimization. Experimental results on two publicly available datasets demonstrate the comparable performance of our model with less help of modal information. Quantitative visualization results further validate the interpretability of our method. The source code will be available at https://github.com/VUT-HFUT/DepthMatters .

Generating image captions in different languages is worth exploring and essential for non-native speakers. Nevertheless, collecting paired annotation for every language is time-consuming and impractical, particularly for minor languages. To this end, the cross-lingual image captioning task is proposed, which leverages existing image-source caption annotation data and wild unrelated target corpus to generate satisfactory caption in the target language. Current methods perform a two-step translation process of image-to-pivot (source) and pivot-to-target. The distinct two-step process comes with certain caption issues, such as the weak semantic alignment between the image and the generated caption and the generated caption’s non-target language style. To address these issues, we propose an end-to-end reinforce learning framework with Visual-linguistic-stylistic Triple Reward named TriR. In TriR, we jointly consider the visual, linguistic, and stylistic alignments to generate factual, fluent, and natural caption in the target language. To be specific, the image-source caption annotation provides factual semantic guidance, whereas the unrelated target corpus guides the language style of generated caption. To achieve this, we construct a visual reward module to measure the cross-modal semantic embedding of image and target caption, a linguistic reward module to measure the cross-linguistic embedding of source and target captions, and a stylistic reward module to imitate the presentation style of target corpus. The TriR can be implemented with either classical CNN-LSTM or prevalent Transformer architecture. Extensive experiments are conducted with four cross-lingual settings, i.e., Chinese-to-English, English-to-Chinese, English-to-German, and English-to-French. Experimental results demonstrate the remarkable superiority of our method, and sufficient ablation experiments validate the beneficial impact of every reward.

Effectively leveraging objects and optical character recognition (OCR) tokens to reason out pivotal scene text is critical for the challenging Text-based Visual Question Answering (TextVQA) task. Graph-based models can effectively capture the semantic relationship among visual entities (objects and tokens) and report remarkable performance in TextVQA. However, previous efforts usually leverage all visual entities and ignore the negative effect of superfluous entities. This article presents a Graph Pooling Inference Network (GPIN), which is an evolutionary graph learning method to purify the visual entities and capture the core semantics. It is observed that the dense distribution of reduplicative objects and the crowd of semantically dependent OCR tokens usually co-exist in the image. Motivated by this, GPIN adopts an adaptive node dropping strategy to dynamically downscale semantically closed nodes for graph evolution and update. To deepen the comprehension of scene text, GPIN is a dual-path hierarchical graph architecture that progressively aggregates the evolved object graph and the evolved token graph semantics into a graph vector that serves as visual cues to facilitate the answer reasoning. It can effectively eliminate object redundancy and enhance the association of semantically continuous tokens. Experiments conducted on TextVQA and ST-VQA datasets show that GPIN achieves promising performance compared with state-of-the-art methods.

Multi-behavioral recommender systems have emerged as a solution to address data sparsity and cold-start issues by incorporating auxiliary behaviors alongside target behaviors. However, existing models struggle to accurately capture varying user preferences across different behaviors and fail to account for diverse item preferences within behaviors. Various user preference factors (such as price or quality) entangled in the behavior may lead to sub-optimization problems. Furthermore, these models overlook the personalized nature of user behavioral preferences by employing uniform transformation networks for all users and items. To tackle these challenges, we propose the Disentangled Cascaded Graph Convolutional Network (Disen-CGCN), a novel multi-behavior recommendation model. Disen-CGCN employs disentangled representation techniques to effectively separate factors within user and item representations, ensuring their independence. In addition, it incorporates a multi-behavioral meta-network, enabling personalized feature transformation across user and item behaviors. Furthermore, an attention mechanism captures user preferences for different item factors within each behavior. By leveraging attention weights, we aggregate user and item embeddings separately for each behavior, computing preference scores that predict overall user preferences for items. Our evaluation of benchmark datasets demonstrates the superiority of Disen-CGCN over state-of-the-art models, showcasing an average performance improvement of 7.07% and 9.00% on respective datasets. These results highlight Disen-CGCN’s ability to effectively leverage multi-behavioral data, leading to more accurate recommendations.

Cognitive diagnosis is a fundamental yet critical research task in the field of intelligent education, which aims to discover the proficiency level of different students on specific knowledge concepts. Despite the effectiveness of existing efforts, previous methods always considered the mastery level on the whole students, so they still suffer from the Long Tail Effect. A large number of students who have sparse interaction records are usually wrongly diagnosed during inference. To relieve the situation, we proposed a Self-supervised Cognitive Diagnosis (SCD) framework which leverages the self-supervised manner to assist the graph-based cognitive diagnosis, then the performance on those students with sparse data can be improved. Specifically, we came up with a graph confusion method that drops edges under some special rules to generate different sparse views of the graph. By maximizing the cross-view consistency of node representations, our model could pay more attention on long-tailed students. Additionally, we proposed an importance-based view generation rule to improve the influence of long-tailed students. Extensive experiments on real-world datasets show the effectiveness of our approach, especially on the students with much sparser interaction records. Our code is available at https://github.com/zeng-zhen/SCD.

Recently, deep Stereo Matching (SM) networks have shown impressive performance and attracted increasing attention in computer vision. However, existing deep SM networks are prone to learn dataset-dependent shortcuts, which fail to generalize well on unseen realistic datasets. This paper takes a step towards training robust models for the domain generalized SM task, which mainly focuses on learning shortcut-invariant representation from synthetic data to alleviate the domain shifts. Specifically, we propose a Hierarchical Visual Transformation (HVT) network to 1) first transform the training sample hierarchically into new domains with diverse distributions from three levels: Global, Local, and Pixel, 2) then maximize the visual discrepancy between the source domain and new domains, and minimize the cross-domain feature inconsistency to capture domain-invariant features. In this way, we can prevent the model from exploiting the artifacts of synthetic stereo images as shortcut features, thereby estimating the disparity maps more effectively based on the learned robust and shortcut-invariant representation. We integrate our proposed HVT network with SOTA SM networks and evaluate its effectiveness on several public SM benchmark datasets. Extensive experiments clearly show that the HVT network can substantially enhance the performance of existing SM networks in synthetic-to-realistic domain generalization.

This paper identifies two kinds of redundancy in the current VideoQA paradigm. Specifically, the current video encoders tend to holistically embed all video clues at different granularities in a hierarchical manner, which inevitably introducesneighboring-frame redundancy that can overwhelm detailed visual clues at the object level. Subsequently, prevailing vision-language fusion designs introduce thecross-modal redundancy by exhaustively fusing all visual elements with question tokens without explicitly differentiating their pairwise vision-language interactions, thus making a pernicious impact on the answering. To this end, we propose a novel transformer-based architecture, that aims to model VideoQA in a redundancy-aware manner. To address the neighboring-frame redundancy, we introduce a video encoder structure that emphasizes the object-level change in neighboring frames, while adopting an out-of-neighboring message-passing scheme that imposes attention only on distant frames. As for the cross-modal redundancy, we equip our fusion module with a novel adaptive sampling, which explicitly differentiates the vision-language interactions by identifying a small subset of visual elements that exclusively support the answer. Upon these advancements, we find this \underlineR edundancy-\underlinea ware trans\underlineformer (RaFormer) can achieve state-of-the-art results on multiple VideoQA benchmarks.

Most efforts in unsupervised domain adaptation (UDA) focus on learning the domain-invariant representations between the two domains. However, such representations may still confuse two patterns due to the domain gap. Considering that semantic information is useful for the final task and domain information always indicates the discrepancy between two domains, to address this issue, we propose to decouple the representations of semantic features from domain features to reduce domain bias. Different from traditional methods, we adopt a simple but effective module with only one domain discriminator to decouple the representations, offering two benefits. Firstly, it eliminates the need for labeled sample pairs, making it more suitable for UDA. Secondly, without adversarial learning, our model can achieve a more stable training phase. Moreover, to further enhance the task-specific features, we employ a causal mechanism to separate semantic features related to causal factors from the overall feature representations. Specially, we utilize a dual-classifier strategy, where each classifier is fed with the entire features and the semantic features, respectively. By minimizing the discrepancy between the outputs of the two classifiers, the causal influence of the semantic features is accentuated. Experiments on several public datasets demonstrate the proposed model can outperform the state-of-the-art methods. Our code is available at: https://github.com/qzxRtY37/DRLC https://github.com/qzxRtY37/DRLC.

Visual Instance Retrieval (VIR) is a hot research topic for its wide application in real world, such as object re-identification in smart city scenarios. However, due to the limited style diversity in source training data, most existing VIR models always fail to generalize well to unseen domain. How to improve the generalizability of VIR models has received increasing attention in most recent years.

Personalized fashion matching task aims to predict the compatible fashion items given available ones for specific users through the effective modeling of the third-order interaction patterns among the user and item pairs. To achieve this, previous methods separately model two key components, user-item and item-item relationships, which ignore the inherent correlations between them and lead to undesirable performance. With a new perspective, this paper proposes to formulate the personalized item matching as the multi-relational connectivity and apply a single-component translation operation to model the targeted third-order interactions. With user-item-item interactions naturally constructing a multi-relational graph, we further device two graph learning modules to enhance the translation-based matching approach from two perspectives,C ontext and Path. The proposed method, named CP-TransMatch, has been tested with extensive experiments on three benchmark fashion datasets and proven effective. It sets the new SOTA for the personalized fashion matching task.

Human life is populated with articulated objects. Current category-level articulated object 9D pose estimation (Articulated Object 9D Pose Estimation, ArtOPE) methods usually meet the challenges of shared object representation requirement, kinematics-agnostic pose modeling and self-occlusions. In this paper, we propose a novel framework called Articulated object 9D Pose Estimation via Reinforcement Learning (ArtPERL), which formulates the category-level ArtOPE as a reinforcement learning problem. Given a point cloud or RGB-D image input, ArtPERL firstly retrieves the part-sensitive articulated object as reference point cloud, and then introduces a joint-centric pose modeling strategy that estimates 9D pose by fitting joint states via reinforced agent training. Finally, we further propose a pose optimization that refine the predicted 9D pose considering kinematic constraints. We evaluate our ArtPERL on various datasets ranging from synthetic point cloud to real-world multi-hinged object. Experiments demonstrate the superior performance and robustness of our ArtPERL. Our work provides a new perspective on category-level articulated object 9D pose estimation and has the potential to be applied in many fields, including robotics, augmented reality, and autonomous driving.

Video moment retrieval (VMR) aims to search for a video segment that matches the search intent in a query sentence, which has received increasing attention in recent years, due to its practical values in various fields. Existing efforts devoted to this interesting yet challenging task typically encode the query sentence and video segments into unstructured global representations for cross-modal interaction and fusion, which may fail to accurately capture the search intent in complex queries with multi-granularity semantics.

Emotional video captioning (EVC) is an emerging task to describe the factual content with the inherent emotion expressed in a video. It is crucial for the EVC task to effectively perceive subtle and ambiguous visual emotion cues in the stage of caption generation. However, existing captioning methods usually overlooked the learning of emotions in user-generated videos, thus making the generated sentence a bit boring and soulless.

Unsupervised pre-training has shown great success in skeleton-based action understanding recently. Existing works typically train separate modality-specific models (i.e., joint, bone, and motion), then integrate the multi-modal information for action understanding by a late-fusion strategy. Although these approaches have achieved significant performance, they suffer from the complex yet redundant multi-stream model designs, each of which is also limited to the fixed input skeleton modality. To alleviate these issues, in this paper, we propose a Unified Multimodal Unsupervised Representation Learning framework, called UmURL, which exploits an efficient early-fusion strategy to jointly encode the multi-modal features in a single-stream manner. Specifically, instead of designing separate modality-specific optimization processes for uni-modal unsupervised learning, we feed different modality inputs into the same stream with an early-fusion strategy to learn their multi-modal features for reducing model complexity. To ensure that the fused multi-modal features do not exhibit modality bias, i.e., being dominated by a certain modality input, we further propose both intra- and inter-modal consistency learning to guarantee that the multi-modal features contain the complete semantics of each modal via feature decomposition and distinct alignment. In this manner, our framework is able to learn the unified representations of uni-modal or multi-modal skeleton input, which is flexible to different kinds of modality input for robust action understanding in practical cases. Extensive experiments conducted on three large-scale datasets, i.e., NTU-60, NTU-120, and PKU-MMD II, demonstrate that UmURL is highly efficient, possessing the approximate complexity with the uni-modal methods, while achieving new state-of-the-art performance across various downstream task scenarios in skeleton-based action representation learning. Our source code is available at https://github.com/HuiGuanLab/UmURL.

With the continuous generation of massive amounts of multimedia data nowadays, hashing has demonstrated significant potentials for large-scale search. To handle the emerging needs for streaming data retrieval, online hashing is drawing more and more attention. For online scenario, data distribution may change and concept drifts may occur as new data is continuously added to the database. Inevitably, hashing models may lose or disrupt the previously obtained knowledge when learning from new information, which is called the problem of catastrophic forgetting. In this paper, we propose a new online hashing method called Self-distillation Dual-memory Online Hashing with Hash Centers, which is abbreviated to SDOH-HC, to overcome this challenge. Specifically, SDOH-HC contains replay and distillation modules. For replay, a dual-memory mechanism is proposed which involves hash centers and exemplars. For knowledge distillation, we let hash centers distill information from themselves, i.e., the version of last round. Additionally, a new objective function is further built on above modules and is solved discretely to learn hash codes. Extensive experiments on three benchmark datasets demonstrate the effectiveness of our method.

Semi-supervised learning (SSL) has been a fundamental challenge in machine learning for decades. The primary family of SSL algorithms, known as pseudo-labeling, involves assigning pseudo-labels to confident unlabeled instances and incorporating them into the training set. Therefore, the selection criteria of confident instances are crucial to the success of SSL. Recently, there has been growing interest in the development of SSL methods that use dynamic or adaptive thresholds. Yet, these methods typically apply the same threshold to all samples, or use class-dependent thresholds for instances belonging to a certain class, while neglecting instance-level information. In this paper, we propose the study of instance-dependent thresholds, which has the highest degree of freedom compared with existing methods. Specifically, we devise a novel instance-dependent threshold function for all unlabeled instances by utilizing their instance-level ambiguity and the instance-dependent error rates of pseudo-labels, so instances that are more likely to have incorrect pseudo-labels will have higher thresholds. Furthermore, we demonstrate that our instance-dependent threshold function provides a bounded probabilistic guarantee for the correctness of the pseudo-labels it assigns.

Attribute-specific fashion retrieval (ASFR) is a challenging information retrieval task, which has attracted increasing attention in recent years. Different from traditional fashion retrieval which mainly focuses on optimizing holistic similarity, the ASFR task concentrates on attribute-specific similarity, resulting in more fine-grained and interpretable retrieval results. As the attribute-specific similarity typically corresponds to the specific subtle regions of images, we propose a Region-to-Patch Framework (RPF) that consists of a region-aware branch and a patch-aware branch to extract fine-grained attribute-related visual features for precise retrieval in a coarse-to-fine manner. In particular, the region-aware branch is first to be utilized to locate the potential regions related to the semantic of the given attribute. Then, considering that the located region is coarse and still contains the background visual contents, the patch-aware branch is proposed to capture patch-wise attribute-related details from the previous amplified region. Such a hybrid architecture strikes a proper balance between region localization and feature extraction. Besides, different from previous works that solely focus on discriminating the attribute-relevant foreground visual features, we argue that the attribute-irrelevant background features are also crucial for distinguishing the detailed visual contexts in a contrastive manner. Therefore, a novel E-InfoNCE loss based on the foreground and background representations is further proposed to improve the discrimination of attribute-specific representation. Extensive experiments on three datasets demonstrate the effectiveness of our proposed framework, and also show a decent generalization of our RPF on out-of-domain fashion images. Our source code is available at https://github.com/HuiGuanLab/RPF.

Text-based visual question answering (TextVQA) faces the significant challenge of avoiding redundant relational inference. To be specific, a large number of detected objects and optical character recognition (OCR) tokens result in rich visual relationships. Existing works take all visual relationships into account for answer prediction. However, there are three observations: (1) a single subject in the images can be easily detected as multiple objects with distinct bounding boxes (considered repetitive objects). The associations between these repetitive objects are superfluous for answer reasoning; (2) two spatially distant OCR tokens detected in the image frequently have weak semantic dependencies for answer reasoning; and (3) the co-existence of nearby objects and tokens may be indicative of important visual cues for predicting answers. Rather than utilizing all of them for answer prediction, we make an effort to identify the most important connections or eliminate redundant ones. We propose a sparse spatial graph network (SSGN) that introduces a spatially aware relation pruning technique to this task. As spatial factors for relation measurement, we employ spatial distance, geometric dimension, overlap area, and DIoU for spatially aware pruning. We consider three visual relationships for graph learning: object-object, OCR-OCR tokens, and object-OCR token relationships. SSGN is a progressive graph learning architecture that verifies the pivotal relations in the correlated object-token sparse graph, and then in the respective object-based sparse graph and token-based sparse graph. Experiment results on TextVQA and ST-VQA datasets demonstrate that SSGN achieves promising performances. And some visualization results further demonstrate the interpretability of our method.

This article tackles the challenging yet important task of Visual Grounding (VG), which aims to localize a visual region in the given image referred by a natural language query. Existing efforts on the VG task are twofold: (1) two-stage methods first extract region proposals and then rank them according to their similarities with the referring expression, which usually leads to suboptimal results due to the quality of region proposals; (2) one-stage methods usually predict all the possible coordinates of the target region online by leveraging modern object detection architectures, which pay little attention to cross-modality correlations and have limited generalization ability. To better address the task, we present an effective transformer-based end-to-end visual grounding approach, which focuses on capturing the cross-modality correlations between the referring expression and visual regions for accurately reasoning the location of the target region. Specifically, our model consists of a feature encoder, a cross-modality interactor, and a modality-agnostic decoder. The feature encoder is employed to capture the intra-modality correlation, which models the linguistic context in query and the spatial dependency in image respectively. The cross-modality interactor endows the model with the capability of highlighting the localization-relevant visual and textual cues by mutual verification of vision and language, which plays a key role in our model. The decoder learns a consolidated token representation enriched by multi-modal contexts and further directly predicts the box coordinates. Extensive experiments on five public benchmark datasets with quantitative and qualitative analysis clearly demonstrate the effectiveness and rationale of our proposed method.

This article targets the task of language-based video moment localization. The language-based setting of this task allows for an open set of target activities, resulting in a large variation of the temporal lengths of video moments. Most existing methods prefer to first sample sufficient candidate moments with various temporal lengths, then match them with the given query to determine the target moment. However, candidate moments generated with a fixed temporal granularity may be suboptimal to handle the large variation in moment lengths. To this end, we propose a novel multi-stage Progressive Localization Network (PLN) that progressively localizes the target moment in a coarse-to-fine manner. Specifically, each stage of PLN has a localization branch and focuses on candidate moments that are generated with a specific temporal granularity. The temporal granularities of candidate moments are different across the stages. Moreover, we devise a conditional feature manipulation module and an upsampling connection to bridge the multiple localization branches. In this fashion, the later stages are able to absorb the previously learned information, thus facilitating the more fine-grained localization. Extensive experiments on three public datasets demonstrate the effectiveness of our proposed PLN for language-based moment localization, especially for localizing short moments in long videos.

Personalized fashion recommendation aims to explore patterns from historical interactions between users and fashion items and thereby predict the future ones. It is challenging due to the sparsity of the interaction data and the diversity of user preference in fashion. To tackle the challenge, this paper investigates multiple factor fields in fashion domain, such as colour, style, brand, and tries to specify the implicit user-item interaction into field level. Specifically, an attentional factor field interaction graph (AFFIG) approach is proposed which models both the user-factor interactions and cross-field factors interactions for predicting the recommendation probability at specific field. In addition, an attention mechanism is equipped to aggregate the cross-field factor interactions for each field. Extensive experiments have been conducted on three E-Commerce fashion datasets and the results demonstrate the effectiveness of the proposed method for fashion recommendation. The influence of various factor fields on recommendation in fashion domain is also discussed through experiments.

Current methods for text-to-video retrieval (T2VR) are trained and tested on video-captioning oriented datasets such as MSVD, MSR-VTT and VATEX. A key property of these datasets is that videos are assumed to be temporally pre-trimmed with short duration, whilst the provided captions well describe the gist of the video content. Consequently, for a given paired video and caption, the video is supposed to be fully relevant to the caption. In reality, however, as queries are not known a priori, pre-trimmed video clips may not contain sufficient content to fully meet the query. This suggests a gap between the literature and the real world. To fill the gap, we propose in this paper a novel T2VR subtask termed Partially Relevant Video Retrieval (PRVR). An untrimmed video is considered to be partially relevant w.r.t. a given textual query if it contains a moment relevant to the query. PRVR aims to retrieve such partially relevant videos from a large collection of untrimmed videos. PRVR differs from single video moment retrieval and video corpus moment retrieval, as the latter two are to retrieve moments rather than untrimmed videos. We formulate PRVR as a multiple instance learning (MIL) problem, where a video is simultaneously viewed as a bag of video clips and a bag of video frames. Clips and frames represent video content at different time scales. We propose a Multi-Scale Similarity Learning (MS-SL) network that jointly learns clip-scale and frame-scale similarities for PRVR. Extensive experiments on three datasets (TVR, ActivityNet Captions, and Charades-STA) demonstrate the viability of the proposed method. We also show that our method can be used for improving video corpus moment retrieval.

This paper attacks the challenging problem of video retrieval by text. In such a retrieval paradigm, an end user searches for unlabeled videos by ad-hoc queries described exclusively in the form of a natural-language sentence, with no visual example provided. Given videos as sequences of frames and queries as sequences of words, an effective sequence-to-sequence cross-modal matching is crucial. To that end, the two modalities need to be first encoded into real-valued vectors and then projected into a common space. In this paper we achieve this by proposing a dual deep encoding network that encodes videos and queries into powerful dense representations of their own. Our novelty is two-fold. First, different from prior art that resorts to a specific single-level encoder, the proposed network performs multi-level encoding that represents the rich content of both modalities in a coarse-to-fine fashion. Second, different from a conventional common space learning algorithm which is either concept based or latent space based, we introduce hybrid space learning which combines the high performance of the latent space and the good interpretability of the concept space. Dual encoding is conceptually simple, practically effective and end-to-end trained with hybrid space learning. Extensive experiments on four challenging video datasets show the viability of the new method. Code and data are available at https://github.com/danieljf24/hybrid_space.

The task of video moment retrieval (VMR) is to retrieve the specific video moment from an untrimmed video, according to a textual query. It is a challenging task that requires effective modeling of complex cross-modal matching relationship. Recent efforts primarily model the cross-modal interactions by hand-crafted network architectures. Despite their effectiveness, they rely heavily on expert experience to select architectures and have numerous hyperparameters that need to be carefully tuned, which significantly limit their applications in real-world scenarios. How to design flexible architectures for modeling cross-modal interactions with less manual effort is crucial for the task of VMR but has received limited attention so far. To address this issue, we present a novel VMR approach that automatically searches for an optimal architecture to learn cross-modal matching relationship. Specifically, we develop a cross-modal architecture searching method. It first searches for repeatable cell network architectures based on a directed acyclic graph, which performs operation sampling over a customized task-specific operation set. Then, we adaptively modulate the edge importance in the graph by a query-aware attention network, which performs edge sampling softly in the searched cell. Different from existing neural architecture search methods, our approach can effectively exploit the query information to reach query-conditioned architectures for modeling cross modal matching. Extensive experiments on three benchmark datasets show that our approach can not only significantly outperform the state-of-the-art methods but also run more efficiently and robustly than manually crafted network architectures.

Conversational systems have recently attracted significant attention. Both the research community and industry believe that it will exert huge impact on human-computer interaction, and specifically, the IR/RecSys community has begun to explore Conversational Recommendation. In real-life scenarios, such systems are often urgently needed in helping users accomplishing different tasks under various situations. However, existing works still face several shortcomings: (1) Most efforts are largely confined in single task setting. They fall short of hands in handling tasks across domains. (2) Aside from soliciting user preference from dialogue history, a conversational recommender naturally has access to the back-end data structure which should be fully leveraged to yield good recommendations. In this paper, we thus present a Topic-guided Conversational Recommender (TCR) which is specifically designed for the multi-domain setting. It augments the sequence-to-sequence (seq2seq) models with a neural latent topic component to better guide the response generation. To better leverage the dialogue history and the back-end data structure, we adopt a graph convolutional network (GCN) to model the relationships between different recommendation candidates while also capture the match between candidates and the dialogue history. We then seamlessly combine these two parts with the idea of pointer networks. We perform extensive evaluation on a large-scale task-oriented multi-domain dialogue dataset and the results show that our method achieves superior performance as compared to a wide range of baselines.

No abstract available.

This companion paper supports the replication of the fashion trend forecasting experiments with the KERN (Knowledge Enhanced Recurrent Network) method that we presented in the ICMR 2020. We provide an artifact that allows the replication of the experiments using a Python implementation. The artifact is easy to deploy with simple installation, training and evaluation. We reproduce the experiments conducted in the original paper and obtain similar performance as previously reported. The replication results of the experiments support the main claims in the original paper.

Video Visual Relation Detection (VidVRD) aims to semantically describe the dynamic interactions across visual concepts localized in a video in the form of subject, predicate, object. It can help to mitigate the semantic gap between vision and language in video understanding, thus receiving increasing attention in multimedia communities. Existing efforts primarily leverage the multimodal/spatio-temporal feature fusion to augment the representation of object trajectories as well as their interactions and formulate the prediction of predicates as a multi-class classification task. Despite their effectiveness, existing models ignore the severe long-tailed bias in VidVRD datasets. As a result, the models' prediction will be easily biased towards the popular head predicates (e.g., next-to and in-front-of), thus leading to poor generalizability.

Deep learning has achieved significant success in multimedia fields involving computer vision, natural language processing, and acoustics. However research in adversarial learning also shows that they are highly vulnerable to adversarial examples. Extensive works have demonstrated that adversarial examples could easily fool deep neural networks to wrong predictions threatening practical deep learning applications in both digital and physical world. Though challenging, discovering and harnessing adversarial attacks is beneficial for diagnosing model blind-spots and further understanding as well as improving multimedia systems in practice. In this workshop, we aim to bring together researchers from the fields of adversarial machine learning, model robustness, and explainable AI to discuss recent research and future directions for adversarial robustness of deep learning models, with a particular focus on multimedia applications, including computer vision, acoustics, etc. As far as we know, we are the first workshop to focus on adversarial learning of multimedia deep learning systems, which is of great significance and we hope will be held annually in conjunction with ACM MM.

Video data are distinct from images for the extra temporal dimension, which results in more content dependencies from various perspectives. It increases the difficulty of learning representation for various video actions. Existing methods mainly focus on the dependency under a specific perspective, which cannot facilitate the categorization of complex video actions. This paper proposes a novel selective dependency aggregation (SDA) module, which adaptively exploits multiple types of video dependencies to refine the features. Specifically, we empirically investigate various long-range and short-range dependencies achieved by the multi-direction multi-scale feature squeeze and the dependency excitation. Query structured attention is then adopted to fuse them selectively, fully considering the diversity of videos' dependency preferences. Moreover, the channel reduction mechanism is involved in SDA for controlling the additional computation cost to be lightweight. Finally, we show that the SDA module can be easily plugged into different backbones to form SDA-Nets and demonstrate its effectiveness, efficiency and robustness by conducting extensive experiments on several video benchmarks for action classification. The code and models will be available at https://github.com/ty-97/SDA.

We tackle the task of video moment retrieval (VMR), which aims to localize a specific moment in a video according to a textual query. Existing methods primarily model the matching relationship between query and moment by complex cross-modal interactions. Despite their effectiveness, current models mostly exploit dataset biases while ignoring the video content, thus leading to poor generalizability. We argue that the issue is caused by the hidden confounder in VMR, i.e., temporal location of moments, that spuriously correlates the model input and prediction. How to design robust matching models against the temporal location biases is crucial but, as far as we know, has not been studied yet for VMR.

This paper strives to predict fine-grained fashion similarity. In this similarity paradigm, one should pay more attention to the similarity in terms of a specific design/attribute between fashion items. For example, whether the collar designs of the two clothes are similar. It has potential value in many fashion related applications, such as fashion copyright protection. To this end, we propose an Attribute-Specific Embedding Network (ASEN) to jointly learn multiple attribute-specific embeddings, thus measure the fine-grained similarity in the corresponding space. The proposed ASEN is comprised of a global branch and a local branch. The global branch takes the whole image as input to extract features from a global perspective, while the local branch takes as input the zoomed-in region-of-interest (RoI) w.r.t. the specified attribute thus able to extract more fine-grained features. As the global branch and the local branch extract the features from different perspectives, they are complementary to each other. Additionally, in each branch, two attention modules, i.e., Attribute-aware Spatial Attention and Attribute-aware Channel Attention, are integrated to make ASEN be able to locate the related regions and capture the essential patterns under the guidance of the specified attribute, thus make the learned attribute-specific embeddings better reflect the fine-grained similarity. Extensive experiments on three fashion-related datasets, i.e., FashionAI, DARN, and DeepFashion, show the effectiveness of ASEN for fine-grained fashion similarity prediction and its potential for fashion reranking. Code and data are available at https://github.com/maryeon/asenpp.

Understanding the mix-and-match relationships between items receives increasing attention in the fashion industry. Existing methods have primarily learned visual compatibility from dyadic co-occurrence or co-purchase information of items to model the item-item matching interaction. Despite effectiveness, rich extra-connectivities between compatible items, e.g., user-item interactions and item-item substitutable relationships, which characterize the structural properties of items, have been largely ignored. This paper presents a graph-based fashion matching framework named Deep Relational Embedding Propagation (DREP), aiming to inject the extra-connectivities between items into the pairwise compatibility modeling. Specifically, we first build a multi-relational item-item-user graph which encodes diverse item-item and user-item relationships. Then we compute structured representations of items by an attentive relational embedding propagation rule that performs messages propagation along edges of the relational graph. This leads to expressive modeling of higher-order connectivity between items and also better representation of fashion items. Finally, we predict pairwise compatibility based on a compatibility metric learning module. Extensive experiments show that DREP can significantly improve the performance of state-of-the-art methods.

Fashion trend forecasting is a crucial task for both academia andindustry. Although some efforts have been devoted to tackling this challenging task, they only studied limited fashion elements with highly seasonal or simple patterns, which could hardly reveal thereal fashion trends. Towards insightful fashion trend forecasting,this work focuses on investigating fine-grained fashion element trends for specific user groups. We first contribute a large-scale fashion trend dataset (FIT) collected from Instagram with extracted time series fashion element records and user information. Furthermore, to effectively model the time series data of fashion elements with rather complex patterns, we propose a Knowledge Enhanced Recurrent Network model (KERN) which takes advantage of the capability of deep recurrent neural networks in modeling time series data. Moreover, it leverages internal and external knowledgein fashion domain that affects the time-series patterns of fashion element trends. Such incorporation of domain knowledge further enhances the deep learning model in capturing the patterns of specific fashion elements and predicting the future trends. Extensive experiments demonstrate that the proposed KERN model can effectively capture the complicated patterns of objective fashion elements, therefore making preferable fashion trend forecast.

Grounding objects in visual context from natural language queries is a crucial yet challenging vision-and-language task, which has gained increasing attention in recent years. Existing work has primarily investigated this task in the context of still images. Despite their effectiveness, these methods cannot be directly migrated into the video context, mainly due to 1) the complex spatio-temporal structure of videos and 2) the scarcity of fine-grained annotations of videos. To effectively ground objects in videos is profoundly more challenging and less explored.

The rapid growth of user-generated videos on the Internet has intensified the need for text-based video retrieval systems. Traditional methods mainly favor the concept-based paradigm on retrieval with simple queries, which are usually ineffective for complex queries that carry far more complex semantics. Recently, embedding-based paradigm has emerged as a popular approach. It aims to map the queries and videos into a shared embedding space where semantically-similar texts and videos are much closer to each other. Despite its simplicity, it forgoes the exploitation of the syntactic structure of text queries, making it suboptimal to model the complex queries.

In Recent years, with the vast development of deep learning techniques, a great deal of effort has been devoted in the computer vision and multimedia community toward the problems of visual object analysis, such as object representation, recognition, detection, identification, etc. Especially at the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) in 2014, the computers have successfully outperformed humans with a lower error rate at image recognition for the first time. However, most of existing algorithms focus more on analyzing objects in a relatively simple and restricted situation, which may perform poorly in natural environments.

Learning effective visual similarity is an essential problem in multimedia research. Despite the promising progress made in recent years, most existing approaches learn visual features and similarities in two separate stages, which inevitably limits their performance. Once useful information has been lost in the feature extraction stage, it can hardly be recovered later. This article proposes a novel end-to-end approach for visual similarity modeling, calleddeep neighborhood component analysis, which discriminatively trains deep neural networks to jointly learn visual features and similarities. Specifically, we first formulate a metric learning objective that maximizes the intra-class correlations and minimizes the inter-class correlations under the neighborhood component analysis criterion, and then train deep convolutional neural networks to learn a nonlinear mapping that projects visual instances from original feature space to a discriminative and neighborhood-structure-preserving embedding space, thus resulting in better performance. We conducted extensive evaluations on several widely used and challenging datasets, and the impressive results demonstrate the effectiveness of our proposed approach.

Identifying mix-and-match relationships between fashion items is an urgent task in a fashion e-commerce recommender system. It will significantly enhance user experience and satisfaction. However, due to the challenges of inferring the rich yet complicated set of compatibility patterns in a large e-commerce corpus of fashion items, this task is still underexplored. Inspired by the recent advances in multirelational knowledge representation learning and deep neural networks, this paper proposes a novel Translation-based Neural Fashion Compatibility Modeling (TransNFCM) framework, which jointly optimizes fashion item embeddings and category-specific complementary relations in a unified space via an end-to-end learning manner. TransNFCM places items in a unified embedding space where a category-specific relation (category-comp-category) is modeled as a vector translation operating on the embeddings of compatible items from the corresponding categories. By this way, we not only capture the specific notion of compatibility conditioned on a specific pair of complementary categories, but also preserve the global notion of compatibility. We also design a deep fashion item encoder which exploits the complementary characteristic of visual and textual features to represent the fashion products. To the best of our knowledge, this is the first work that uses category-specific complementary relations to model the category-aware compatibility between items in a translation-based embedding space. Extensive experiments demonstrate the effectiveness of TransNFCM over the state-of-the-arts on two real-world datasets.

Most existing image enhancement methods function like a black box, which cannot clearly reveal the procedure behind each image enhancement operation. To overcome this limitation, in this paper, we design a progressive image enhancement framework, which generates an expected "good" retouched image with a group of self-interpretable image filters under the guidance of an aesthetic assessment model. The introduced aesthetic network effectively alleviates the shortage of paired training samples by providing extra supervision, and eliminate the bias caused by human subjective preferences. The self-interpretable image filters designed in our image enhancement framework, make the overall image enhancing procedure easy-to-understand. Extensive experiments demonstrate the effectiveness of our proposed framework.

With the rapid evolution of social networks, the increasing user intention gap and visual semantic gap both bring great challenge for users to access satisfied contents. It becomes promising to investigate users' customized multimedia recommendation. In this paper, we propose cross-modal collaborative manifold propagation (CMP) for image recommendation. CMP leverages users' interest distribution to propagate images' user records, which lets users know the trend from others and produces interest-aware image candidates upon users' interests. Visual distribution is investigated simultaneously to propagate users' visual records along dense semantic visual manifold. Visual manifold propagation helps to estimate semantic accurate user-image correlations for the candidate images in recommendation ranking. Experimental performance demonstrate the collaborative user-image inferring ability of CMP with effective user interest manifold propagation and semantic visual manifold propagation in personalized image recommendation.

Understanding the objects and relations between them is indispensable to fine-grained video content analysis, which is widely studied in recent research works in multimedia and computer vision. However, existing works are limited to evaluating with either small datasets or indirect metrics, such as the performance over images. The underlying reason is that the construction of a large-scale video dataset with dense annotation is tricky and costly. In this paper, we address several main issues in annotating objects and relations in user-generated videos, and propose an annotation pipeline that can be executed at a modest cost. As a result, we present a new dataset, named VidOR, consisting of 10k videos (84 hours) together with dense annotations that localize 80 categories of objects and 50 categories of predicates in each video. We have made the training and validation set public and extendable for more tasks to facilitate future research on video object and relation recognition.

Fashion knowledge helps people to dress properly and addresses not only physiological needs of users, but also the demands of social activities and conventions. It usually involves three mutually related aspects of: occasion, person and clothing. However, there are few works focusing on extracting such knowledge, which will greatly benefit many downstream applications, such as fashion recommendation. In this paper, we propose a novel method to automatically harvest fashion knowledge from social media. We unify three tasks of occasion, person and clothing discovery from multiple modalities of images, texts and metadata. For person detection and analysis, we use the off-the-shelf tools due to their flexibility and satisfactory performance. For clothing recognition and occasion prediction, we unify the two tasks by using a contextualized fashion concept learning module, which captures the dependencies and correlations among different fashion concepts. To alleviate the heavy burden of human annotations, we introduce a weak label modeling module which can effectively exploit machine-labeled data, a complementary of clean data. In experiments, we contribute a benchmark dataset and conduct extensive experiments from both quantitative and qualitative perspectives. The results demonstrate the effectiveness of our model in fashion concept prediction, and the usefulness of extracted knowledge with comprehensive analysis.

Food recognition for user-uploaded images is crucial in visual diet tracking, an emerging application linking multimedia and healthcare domains. However, it is challenging due to the various visual appearances of food images. This is caused by different conditions when taking the photos, such as angles, distances, light conditions, food containers, and background scenes. To alleviate such a semantic gap, this paper presents a cross-modal alignment and transfer network (ATNet), which is motivated by the paradigm of learning using privileged information (LUPI). It additionally utilizes the ingredients in food images as an "intelligent teacher" in the training stage to facilitate cross-modal information passing. Specifically, ATNet first uses a pair of synchronized autoencoders to build the base image and ingredient channels for information flow. Subsequently, the information passing is enabled through a two-stage cross-modal interaction. The first stage of interaction adopts a two-step method, called partial heterogeneous transfer, to 1) alleviate the intrinsic heterogeneity between images and ingredients and 2) align them in a shared space to make their carried information about food classes interact. In the second stage, ATNet learns to map the visual embeddings of images to the ingredient channel for food recognition from the view of "teacher''. This leads a refined recognition by a multi-view fusion. Experiments on two real-world datasets show that ATNet can be incorporated with any state-of-the-art CNN models to consistently improve their performance.

Understanding the mix-and-match relationships of fashion items receives increasing attention in fashion industry. Existing methods have primarily utilized the visual content to learn the visual compatibility and performed matching in a latent space. Despite their effectiveness, these methods work like a black box and cannot reveal the reasons that two items match well. The rich attributes associated with fashion items, e.g.,off-shoulder dress and black skinny jean, which describe the semantics of items in a human-interpretable way, have largely been ignored.

Despite the promising progress made in recent years, person reidentification (re-ID) remains a challenging task due to the complex variations in human appearances from different camera views. This paper proposes to tackle this task by jointly learning feature representation and distance metric in an end-to-end manner. Existing deep metric learning-based re-ID methods usually encounter the following two weaknesses: 1) most works based on pairwise or triplet constraints often suffer from slow convergence and poor local optima, partially because they use very limited samples for each update and 2) hard negative sample mining has been widely applied in existing works. However, hard positive samples, which also contribute to the training of network, have not received enough attention. To alleviate these problems, we develop a novel structural metric learning objective for person re-ID, in which each positive pair is allowed to be compared against all negative pairs in a minibatch and each positive pair is adaptively assigned a hardness-aware weight to modulate its contribution. The introduced positive pair weighting strategy enables the algorithm to focus more on the hard positive samples. Furthermore, we propose to enhance the proposed loss function by adding a global loss term to reduce the variances of positive/negative pair distances, which is able to improve the generalization capability of the network model. By this approach, person images can be nonlinearly mapped into a low-dimensional embedding space where similar samples are kept closer and dissimilar samples are pushed farther apart. We implement the proposed algorithm using the inception architecture and evaluate it on three large-scale re-ID data sets. Experiment results demonstrate that our approach is able to outperform most state of the arts while using much lower dimensional deep features.

Despite the promising progress made in recent years, person re-identification remains a challenging task due to complex variations in human appearances from different camera views. This paper presents a logistic discriminant metric learning method for this challenging problem. Different with most existing metric learning algorithms, it exploits both original data and auxiliary data during training, which is motivated by the new machine learning paradigm-learning using privileged information. Such privileged information is a kind of auxiliary knowledge, which is only available during training. Our goal is to learn an optimal distance function by constructing a locally adaptive decision rule with the help of privileged information. We jointly learn two distance metrics by minimizing the empirical loss penalizing the difference between the distance in the original space and that in the privileged space. In our setting, the distance in the privileged space functions as a local decision threshold, which guides the decision making in the original space like a teacher. The metric learned from the original space is used to compute the distance between a probe image and a gallery image during testing. In addition, we extend the proposed approach to a multi-view setting which is able to explore the complementation of multiple feature representations. In the multi-view setting, multiple metrics corresponding to different original features are jointly learned, guided by the same privileged information. Besides, an effective iterative optimization scheme is introduced to simultaneously optimize the metrics and the assigned metric weights. Experiment results on several widely-used data sets demonstrate that the proposed approach is superior to global decision threshold-based methods and outperforms most state-of-the-art results.

Despite the promising progress made in recent years, person re-identification (re-ID) remains a challenging task due to the complex variations in human appearances from different camera views. For this challenging problem, a large variety of algorithms have been developed in the fully supervised setting, requiring access to a large amount of labeled training data. However, the main bottleneck for fully supervised re-ID is the limited availability of labeled training samples. To address this problem, we propose a self-trained subspace learning paradigm for person re-ID that effectively utilizes both labeled and unlabeled data to learn a discriminative subspace where person images across disjoint camera views can be easily matched. The proposed approach first constructs pseudo-pairwise relationships among unlabeled persons using the k-nearest neighbors algorithm. Then, with the pseudo-pairwise relationships, the unlabeled samples can be easily combined with the labeled samples to learn a discriminative projection by solving an eigenvalue problem. In addition, we refine the pseudo-pairwise relationships iteratively, which further improves learning performance. A multi-kernel embedding strategy is also incorporated into the proposed approach to cope with the non-linearity in a person’s appearance and explore the complementation of multiple kernels. In this way, the performance of person re-ID can be greatly enhanced when training data are insufficient. Experimental results on six widely used datasets demonstrate the effectiveness of our approach, and its performance can be comparable to the reported results of most state-of-the-art fully supervised methods while using much fewer labeled data.

Saliency detection has recently received increasing research interest on using high-dimensional datasets beyond two-dimensional images. Despite the many available capturing devices and algorithms, there still exists a wide spectrum of challenges that need to be addressed to achieve accurate saliency detection. Inspired by the success of the light-field technique, in this article, we propose a new computational scheme to detect salient regions by integrating multiple visual cues from light-field images. First, saliency prior maps are generated from several light-field features based on superpixel-level intra-cue distinctiveness, such as color, depth, and flow inherited from different focal planes and multiple viewpoints. Then, we introduce the location prior to enhance the saliency maps. These maps will finally be merged into a single map using a random-search-based weighting strategy. Besides, we refine the object details by employing a two-stage saliency refinement to obtain the final saliency map. In addition, we present a more challenging benchmark dataset for light-field saliency analysis, named HFUT-Lytro , which consists of 255 light fields with a range from 53 to 64 images generated from each light-field image, therein spanning multiple occurrences of saliency detection challenges such as occlusions, cluttered background, and appearance changes. Experimental results show that our approach can achieve 0.6--6.7% relative improvements over state-of-the-art methods in terms of the F-measure and Precision metrics, which demonstrates the effectiveness of the proposed approach.