Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks94 from public data
- Recurrent Squeeze-and-Excitation Context Aggregation Net for Single Image Deraining940
A novel deep network architecture based on deep convolutional and recurrent neural networks for single image deraining based on contextual information is proposed and outperforms the state-of-the-art approaches under all evaluation metrics.
- Expectation-Maximization Attention Networks for Semantic Segmentation663
This paper forms the attention mechanism into an expectation-maximization manner and iteratively estimate a much more compact set of bases upon which the attention maps are computed, which is robust to the variance of input and is also friendly in memory and computation.
- Video parsing, retrieval and browsing400
This paper presents an integrated solution for computer assisted video parsing and content-based video retrieval and browsing that uses video content information provided by a parsing process driven by visual feature analysis.
- 306
- Dynamic Modality Interaction Modeling for Image-Text Retrieval230
A novel modality interaction modeling network based upon the routing mechanism, which is the first unified and dynamic multimodal interaction framework towards image-text retrieval and demonstrates superiority compared with several state-of-the-art baselines.
- Deep Comprehensive Correlation Mining for Image Clustering204
A novel clustering framework, named deep comprehensive correlation mining~(DCCM), for exploring and taking full advantage of various kinds of correlations behind the unlabeled data from three aspects: Instead of only using pair-wise information, pseudo-label supervision is proposed to investigate category information and learn discriminative features.
- Agree to Disagree: Adaptive Ensemble Knowledge Distillation in Gradient Space144
This paper examines the diversity of teacher models in the gradient space and regard the ensemble knowledge distillation as a multi-objective optimization problem so that the method can be seen as a dynamic weighting method for each teacher in the ensemble.
- Stacked Hybrid-Attention and Group Collaborative Learning for Unbiased Scene Graph Generation135
A novel Stacked Hybrid-Attention network is presented, which facilitates the intra-modal refinement as well as the intermodal interaction, to serve as the encoder and an innovative Group Collaborative Learning strategy is devised to optimize the decoder.
- Overcoming cold tumors: a combination strategy of immune checkpoint inhibitors113
Understanding the transformation from ‘cold’ to ‘hot’ tumors is essential in developing effective cancer treatments, and identifying key factors that impact tumor T cell infiltration is vital for creating effective treatments targeting ‘cold’ tumors.
- Unified Graph and Low-Rank Tensor Learning for Multi-View Clustering110
This paper proposes the unified graph and low-rank tensor learning (UGLTL) for multi-view clustering and unifies the view-specific affinity matrix based on projected graph learning and reorganizes the affinity matrices into tensor form and learns its intrinsic tensor based on low- rank tensor approximation.
- Detecting and Grounding Multi-Modal Media Manipulation and Beyond98
This paper aims to not only detect the authenticity of multi-modal media, but also ground the manipulated content (i.e., image bounding boxes and text tokens), which requires deeper reasoning of multi-modal media manipulation.
- WKVQuant: Quantizing Weight and Key/Value Cache for Large Language Models Gains More88
This paper proposes WKVQuant, a PTQ framework especially designed for quantizing weights and the key/value (KV) cache of LLMs that incorporates past-only quantization to improve the computation of attention and introduces two-dimensional quantization strategy to handle the distribution of KV cache.
- 75
- SOGNet: Scene Overlap Graph Network for Panoptic Segmentation65
This study aims to model overlap relations among instances and resolve them for panoptic segmentation, Inspired by scene graph representation, and forms the overlapping problem as a simplified case, named scene overlap graph.
- High Quality Segmentation for Ultra High-resolution Images63
The Continuous Refinement Model (CRM) continuously aligns the feature map with the refinement target and aggregates features to reconstruct these image details and shows its significant generalization ability to fill the resolution gap between low-resolution training images and ultra high-resolution testing ones.
- Local Correlation Consistency for Knowledge Distillation60
To make the student focus on those informative local regions of the teacher's feature maps, a novel class-aware attention module is proposed to highlight the class-relevant regions and remove the confusing class-irrelevant regions, which makes the local correlation knowledge more accurate and valuable.
- 57
- AffordGrasp: In-Context Affordance Reasoning for Open-Vocabulary Task-Oriented Grasping in Clutter45
AffordGrasp is a novel open-vocabulary grasping framework that leverages the reasoning capabilities of vision-language models (VLMs) for in-context affordance reasoning and infers tasks directly from implicit user instructions, enabling more intuitive and seamless human-robot interaction in everyday scenarios.
- Self-Training Boosted Multi-Factor Matching Network for Composed Image Retrieval41
This work designs an iterative dual self-training paradigm to further enhance the performance of LIMN by fully utilizing the potential unlabeled reference-target image pairs in a weakly-supervised manner and denotes the iterative dual self-training paradigm enhanced LIMN as LIMN+.
- Joint Latent Subspace Learning and Regression for Cross-Modal Retrieval41
This paper first adopts the spectral regression method to learn the optimal latent space shared by data of all modalities based on the orthogonal constraints, then builds a graph model to project the multi-modality data into the latent space.
- Multi-modal cooking workflow construction for food recipes39
MM-ReS is built, the first large-scale dataset for cooking workflow construction, consisting of 9,850 recipes with human-labeled workflow graphs and a neural encoder-decoder model that utilizes both visual and textual information to construct the cooking workflow, which achieved over 20% performance gain over existing hand-crafted baselines.
- Multiple Models Fusion for Emotion Recognition in the Wild37
This paper proposes a multiple models fusion method to automatically recognize the expression in the video clip as part of the third Emotion Recognition in the Wild Challenge (EmotiW 2015).
- 36
- A Survey on Video Temporal Grounding With Multimodal Large Language Model34
Current research on VTG-MLLMs is examined through a three-dimensional taxonomy: the functional roles of MLLMs, highlighting their architectural significance; training paradigms, analyzing strategies for temporal reasoning and task adaptation; and video feature processing techniques, which determine spatiotemporal representation effectiveness.
- Fashion Compatibility Modeling through a Multi-modal Try-on-guided Scheme31
A new compatibility modeling scheme which integrates the outfit try-on appearance into the traditional discrete item interaction modeling, and is demonstrated to have superiority to state-of-the-arts modeling.
- Locality-constrained linear coding based bi-layer model for multi-view facial expression recognition30
This paper proposes a simple yet effective method, called the locality-constrained linear coding based bi-layer (LLCBL) model, to learn discriminative representation for multi-view facial expression recognition that outperforms the state-of-the-art methods.
- 29
- Visual Perturbation-aware Collaborative Learning for Overcoming the Language Prior Problem29
This paper proposes a novel collaborative learning scheme from the viewpoint of visual perturbation calibration, which can better investigate the fine-grained visual effects and mitigate the language prior problem by learning the instance-level characteristics.
- 25
- 25
- Dynamical System Inspired Adaptive Time Stepping Controller for Residual Network Families24
An adaptive time stepping controller is developed that is dependent on the parameters of the current step, and aware of previous steps, and jointly optimized with the network training so that variable step sizes and evolution time can be adaptively adjusted.
- 22
- 22
- 21
- 21
- Differential-Perceptive and Retrieval-Augmented MLLM for Change Captioning20
This work aims to empower MLLMs with the capability to perceive subtle differences between paired images and enhance their performance in generating change captions by presenting a diFferental-perceptive, medieval-augmented MLLM (FINER-MLLM) tailored for this task.
- Neural Collapse Terminus: A Unified Solution for Class Incremental Learning and Its Variants18
A unified solution to the misalignment dilemma in the three tasks of CIL, LTCIL and FSCIL is offered andoretical analysis indicates that the method holds the neural collapse optimality in an incremental fashion regardless of data imbalance or data scarcity.
- 17
- Multi-view common space learning for emotion recognition in the wild17
This paper proposes a method to analyse multi-view features of emotion samples and automatically recognize the expression as part of the fourth Emotion Recognition in the Wild Challenge (EmotiW 2016).
- 16
- Semantic-Aware Modular Capsule Routing for Visual Question Answering16
A Semantic-aware modUlar caPsulE Routing framework, termed as SUPER, is proposed to better capture the instance-specific vision-semantic characteristics and refine the discriminative representations for prediction to provide a novel perspective towards architecture learning and representation calibration for VQA.
- Mamba-FSCIL: Dynamic Adaptation With Selective State Space Model for Few-Shot Class Incremental Learning15
This study designs a dual selective SSM projector that generates input-conditioned state-space parameters from intermediate features for dynamic adaptation and develops a class-sensitive selective scan mechanism to guide dynamic adaptation of the incremental branch.
- LipGen: Viseme-Guided Lip Video Generation for Enhancing Visual Speech Recognition14
A novel framework, LipGen, is proposed, which aims to improve model robustness by leveraging speech-driven synthetic visual data, thereby mitigating the constraints of current datasets and introduces an auxiliary task that incorporates viseme classification alongside attention mechanisms.
- 14
- Temporal Sentence Grounding in Streaming Videos13
Two novel methods are proposed: a TwinNet structure that enables the model to learn about upcoming events; and a language-guided feature compressor that eliminates redundant visual frames and reinforces the frames that are relevant to the query.
- R$$^{2}$$-Net: Recurrent and Recursive Network for Sparse-View CT Artifacts Removal13
A novel neural network architecture to reduce streak artifacts generated in sparse-view 2D Computed Tomography image reconstruction is proposed, which outperforms the state-of-the-art methods both quantitatively and qualitatively.
- FineBadminton: A Multi-Level Dataset for Fine-Grained Badminton Video Understanding12
This work introduces FineBadminton, a novel and large-scale dataset featuring a unique multi-level semantic annotation hierarchy (Foundational Actions, Tactical Semantics, and Decision Evaluation) for comprehensive badminton understanding and proposes an optimized baseline approach incorporating Hit-Centric Keyframe Selection to focus on pivotal moments and Coordinate-Guided Condensation to distill salient visual information.
- Matrix recovery with implicitly low-rank data12
A novel method is proposed for matrix recovery that pursues the low-rank structure of the target matrix in an implicit feature space by making use of the specifics of an accelerated proximal gradient based optimization algorithm.
- Micro-video Tagging via Jointly Modeling Social Influence and Tag Relation11
This work forms micro-video tagging as a link prediction problem in a constructed heterogeneous network and combines tag relation, video-tag annotation, and user follow relation to build the network.
- 11
- Mask Again: Masked Knowledge Distillation for Masked Video Modeling10
Extensive experimental results show that MaskAgain achieves comparable or even better performance than existing methods on benchmark datasets with much fewer training epochs and much less memory, which demonstrates thatMaskAgain allows for efficient pre-training of accurate video models, reducing computational resources and training time significantly.
- Joint Dictionary Learning and Semantic Constrained Latent Subspace Projection for Cross-Modal Retrieval10
A novel joint dictionary learning and semantic constrained latent subspace learning method for cross-modal retrieval~(JDSLC) to deal with above two issues and outperforms the state-of-the-art approaches.
- 9
- Fine-grained Key-Value Memory Enhanced Predictor for Video Representation Learning8
This work proposes a novel memory-enhanced predictor that leverages key-value memory networks with separate memories for the online and target encoders, and introduces a visual concept alignment module that ensures fine-grained alignment of shared semantic information across segments of the same video.
- 8
- 7
- SNP-S3: Shared Network Pre-Training and Significant Semantic Strengthening for Various Video-Text Tasks7
The Significant Semantic Strengthening (S3) strategy, which includes a novel masking and matching proxy task to promote the pre-training performance, is proposed, which establishes a new state-of-the-art in pixel-level video-text pre-training.
- PCDVQ: Enhancing Vector Quantization for Large Language Models via Polar Coordinate Decoupling6
Polar Coordinate Decoupled Vector Quantization (PCDVQ), an effective and efficient VQ framework consisting of two key modules: 1) Polar Coordinate Decoupling (PCD), which transforms vectors into their polar coordinate representations and perform independent quantization of the direction and magnitude parameters.
- 6
- Ingredient-enriched Recipe Generation from Cooking Videos5
A framework that introduces ingredient recognition module which uses the copy mechanism to fuse the predicted ingredient information into the generated sentence is proposed which achieves the promising performances on both YouCookII and Cooking-COIN datasets.
- MegaSR: Mining Customized Semantics and Expressive Guidance for Real-World Image Super-Resolution4
This work proposes MegaSR, a novel framework that mines customized block-wise semantics and expressive guidance for diffusion-based ISR, and enables flexible adaptation to multi-granularity semantic awareness by dynamically incorporating image attributes at each block.
- SSCNet: learning-based subspace clustering4
This work redefine and generalize the optimization procedure of the linearized alternating direction method of multipliers (ADMM), framing it as a multi-block deep neural network, where each block corresponds to a linearized ADMM iteration step, to address the SSC problem.
- 3
- OFAR: A Multimodal Evidence Retrieval Framework for Illegal Live-streaming Identification3
A multimodal evidence retrieval system, named OFAR, to facilitate the illegal live-streaming identification, which achieves significant improvement on the industrial dataset TaoLive, demonstrating the advances of the scheme.
- 3
- Gaze‐matching eyebox expansion in retinal projection augmented reality display2
Both simulation and optical experiments show that the proposed eyebox‐extended retinal projection display with gaze‐matching for augmented reality has the ability to project virtual images onto the fovea of the human eye to achieve gaze‐matching.
- Preview-Based Category Contrastive Learning for Knowledge Distillation2
This work proposes a novel preview-based category contrastive learning method for knowledge distillation (PCKD), which first distills the structural knowledge of both instance-level feature correspondence and the relation between instance features and category centers in a contrastive learning fashion, which can explicitly optimize the category representation.
- 2
- 2
- 2
- 2
- 2
- 1
- 1
- Quantitative contrast-enhanced ultrasonography in the diagnosis and grading of hepatic steatosis in brain-dead donors1
The PIHA, a CEUS-derived parameter, could serve as a diagnostic tool for moderate-to-severe HS, and showed promising results in the diagnosis and grading of HS in brain-dead donors.
- 1
- 1
- 1
- –
- –
- –
- GenView++: Unifying Adaptive Generative Augmentation and Quality-Driven Supervision for Contrastive Representation Learning–
The proposed GenView++ introduces a multi-source adaptive view generation mechanism to synthesize diverse yet semantically coherent views by dynamically modulating generative parameters across image-conditioned, text-conditioned, and image-text-conditioned strategies.
- –
- –
- –
- Expert consensus on the off-label use in China of drugs for rare hematologic diseases (2024 edition)–
The consensus aims to provide a framework and reference for the clinical application of off-label drug use in treating rare hematologic diseases and is derived from discussions between Guangdong Pharmaceutical Association Hematologic Rare Diseases Group experts.
- –
- –
- –
- Study on Patent Strategy of High a New Technology Enterprises Cluster–
The paper designs influencing factors analysis framework for cluster patent strategy, and then builds promoting process of clusters patent strategy around patent development, application, use and protection so as to provide theory and method support for patent strategy management of high & new technology enterprises cluster.
- Study on the Performance Evaluation System of Patent Strategy of High a New Technology Enterprise Clusters–
The article designed the cluster patent strategy performance evaluation index system and constructed the performance evaluation method by using anal ytic hierarchy process and fuzzy comprehensive evaluation comprehensively to provide method supporting for patent strategy management of regional high-tech enterprise clusters.
- –
- –
- Optical flow estimation through velocity-based surface fitting–
The velocity-based technique adapts the degree of smoothing by changing the size of the fitting window based on the calculated motion field, and therefore produces a more accurate optical flow.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.