Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Xindong WuSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works219 from public data
- Hierarchical LSTM for Sign Language Translation216
A hierarchical-LSTM (HLSTM) encoder-decoder model with visual content and word embedding for SLT exhibits promising performance on singer-independent test with seen sentences and also outperforms the comparison algorithms on unseen sentences.
- Benchmarking Micro-Action Recognition: Dataset, Methods, and Applications175
A new micro-action dataset designated as Micro-action-52 (MA-52), and a benchmark named micro-action network (MANet) for micro-action recognition (MAR) task are proposed, which assesses MANet and other nine prevalent action recognition methods.
- DADNet174
A novel deep model called Dilated-Attention-Deformable ConvNet (DADNet), which consists of two schemes: multi-scale dilated attention and deformable convolutional DME (Density Map Estimation).
- Proposal-Free Video Grounding with Contextual Pyramid Network148
This paper proposes a novel proposal-free framework named Contextual Pyramid Network (CPNet) to investigate multi-scale temporal correlation in the video and implements a temporal self-attentive regression to predict the activity boundary from the above hierarchical context-aware 2D correlation maps.
- 130
- 117
- Contrastive Positive Sample Propagation Along the Audio-Visual Event Line115
A new contrastive positive sample propagation (CPSP) method is proposed for better deep feature representation learning and is more helpful to obtain the refined audio-visual features that are distinguishable from the negatives.
- Connectionist Temporal Fusion for Sign Language Translation108
A hybrid deep architecture which consists of a temporal convolution module (TCOV), a bidirectional gated recurrent unit module (BGRU), and a fusion layer module (FL) to address the CSLT problem and achieves comparable performance to other existing methods with multiple EM iterations.
- 107
- Audio-Visual Segmentation with Semantics104
A new baseline method is proposed that uses a temporal pixel-wise audio-visual interaction module to inject audio semantics as guidance for the visual segmentation process, and a regularization loss is designed to encourage audio-visual mapping during training.
- Dense Temporal Convolution Network for Sign Language Translation99
A dense temporal convolution network, termed DenseTCN which captures the actions in hierarchical views and addresses the SLT problem by different views, including embedded short-term and extended longterm sequential learning.
- MEGCF: Multimodal Entity Graph Collaborative Filtering for Personalized Recommendation83
A novel model, multimodal entity graph collaborative filtering, short for MEGCF, is developed that captures the semantic correlation between interactions and the features obtained from MFE, thus making a better match between MFE and UIM.
- ViGT: proposal-free video grounding with a learnable token in the transformer79
A novel boundary regression paradigm that performs regression token learning in a transformer, namely video grounding transformer (ViGT), which predicts the temporal boundary using a learnable regression token rather than multi- modal or cross-modal features.
- 79
- Prototypical Calibrating Ambiguous Samples for Micro-Action Recognition76
A novel Prototypical Calibrating Ambiguous Network (PCAN) is proposed to unleash and mitigate the ambiguity of MAR and proposes a prototype-guided rectification to rectify prediction by incorporating the representability of prototypes.
- Transformer-Based Visual Grounding with Cross-Modality Interaction75
This article presents 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.
- 71
- PsycoLLM: Enhancing LLM for Psychological Understanding and Evaluation68
This article proposes a specialized psychological LLM, named PsycoLLM, trained on a proposed high-quality psychological dataset, including single-turn QA, multiturn dialogues, and knowledge-based QA, and develops a comprehensive psychological benchmark based on authoritative psychological counseling examinations in China.
- Gloss Semantic-Enhanced Network with Online Back-Translation for Sign Language Production68
A Gloss semantic-Enhanced Network is proposed with Online Back-Translation (GEN-OBT) for G2P in the SLP task, which outperforms the state-of-the-art models and designs a CTC-based reverse decoder to convert the generated poses backward into glosses, which guarantees the semantic consistency during the processes of gloss- to-pose and pose-to-gloss.
- 65
- Online Early-Late Fusion Based on Adaptive HMM for Sign Language Recognition65
This work proposes an adaptive HMM method to obtain the hidden state number of each sign by affinity propagation clustering and discovers that inherent latent change states of eachsign are related not only to the number of key gestures and body poses but also to their translation relationships.
- EgoTextVQA: Towards Egocentric Scene-Text Aware Video Question Answering61
This work introduces EgoTextVQA, a novel and rigorously constructed benchmark for egocentric QA assistance involving scene text, and comprehensively evaluates 10 prominent multimodal large language models.
- Iterative Context-Aware Graph Inference for Visual Dialog61
This work proposes a novel Context-Aware Graph (CAG) neural network, where each node in the graph corresponds to a joint semantic feature, including both object-based (visual) and history-related (textual) context representations.
- Joint Multi-Grained Popularity-Aware Graph Convolution Collaborative Filtering for Recommendation57
This work develops a Joint Multigrained Popularity-aware Graph Convolution Collaborative Filtering model, short for JMP-GCF, which uses a popularity-aware embedding generation to construct multigrained popularity features and uses the idea of joint learning to capture the signals within and between different granularities of popularity features that are relevant for modeling user preferences.
- Connectionist Temporal Modeling of Video and Language: a Joint Model for Translation and Sign Labeling57
A Connectionist Temporal Modeling (CTM) network for sentence translation and sign labeling is proposed and dynamic programming is embedded into the decoding scheme, which learns temporal mapping among features, sign labels, and the generated sentence directly.
- Repetitive Action Counting With Hybrid Temporal Relation Modeling50
A novel method named Hybrid Temporal Relation Modeling Network (HTRM-Net) is proposed to build diverse TSSM for RAC, which surpasses the classical TransRAC method by 20.04% in MAE and 22.76% in OBO.
- Frequency Decoupling for Motion Magnification Via Multi-Level Isomorphic Architecture49
Inspired by the frequency spectrum, a new paradigm of Frequency Decoupling for Motion Magnification with a Multi-level Isomorphic Architecture to capture multi-level high-frequency details and a stable low-frequency structure (motion field) in video space is presented.
- Advancing Weakly-Supervised Audio-Visual Video Parsing via Segment-Wise Pseudo Labeling49
A new pseudo label generation strategy that can explicitly assign labels to each video segment by utilizing prior knowledge learned from the open world is proposed and achieves state-of-the-art video parsing performance on all types of event parsing, i.e., audio event, visual event, and audio-visual event.
- 47
- Recurrent Relational Memory Network for Unsupervised Image Captioning45
This paper proposes a novel memory-based network rather than GAN, named Recurrent Relational Memory Network (R2M), which implements a concepts-to-sentence memory translator through two-stage memory mechanisms: fusion and recurrent memories.
- 41
- 41
- Dual Visual Attention Network for Visual Dialog41
A novel Dual Visual Attention Network (DVAN) is proposed to realize cross-modal semantic correlation for visual dialog, where Vg, Vl, Q, H and H have inseparable relevances and are mutually embedded to learn the complementary of visual semantics.
- Emotion-Prior Awareness Network for Emotional Video Captioning39
This paper proposes a new emotional captioning perspective in a human-like perception-priority manner, i.e., first perceiving the inherent emotion and then leveraging the perceived emotion cue to support caption generation, and devise an Emotion-Prior Awareness Network (EPAN), which mainly benefits from a novel tree-structured emotion learning module.
- 38
- Data Augmentation for Human Behavior Analysis in Multi-Person Conversations38
This paper presents the solution of the team HFUT-VUT for the MultiMediate Grand Challenge 2023 at ACM Multimedia 2023, selecting Swin Transformer as the baseline and exploiting data augmentation strategies to address the above three tasks.
- Dual-Path TokenLearner for Remote Photoplethysmography-Based Physiological Measurement With Facial Videos36
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.
- 34
- 32
- Pattern matching with wildcards and gap-length constraints based on a centrality-degree graph32
This paper proposes a graph structure WON-Net and designs the WOW (WOW stands for pattern matching with wildcards based on WON-Net) algorithm with the weighted centralization measure based on nodes’ centrality-degrees and defines a new variant of WOW as WOW-δ.
- Discrete to Continuous: Generating Smooth Transition Poses from Sign Language Observations31
This work proposes a novel framework, Sign-D2C, that employs a conditional diffusion model to synthesize contextually smooth transition frames, enabling the seamless construction of continuous sign language sequences.
- Towards Efficient Partially Relevant Video Retrieval With Active Moment Discovering31
This work is committed to discovering video moments that are semantically consistent with their queries and introduces 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.
- Exploring Sparse Spatial Relation in Graph Inference for Text-Based VQA30
A sparse spatial graph network (SSGN) is proposed that introduces a spatially aware relation pruning technique to this task and considers three visual relationships for graph learning: object-object, OCR-OCR tokens, and object-OCR token relationships.
- Clinical photoacoustic/ultrasound dual-modal imaging: Current status and future trends30
The current PAT/US dual-modal imaging systems are discussed in detail, and their promising clinical applications are presented and compared systematically and the potential impacts of these combined systems in the coming future are described.
- Learning from Heterogeneity: Generalizing Dynamic Facial Expression Recognition via Distributionally Robust Optimization29
A novel framework, called Heterogeneity-aware Distributional Framework (HDF), and two plug-and-play modules to enhance time-frequency modeling and mitigate optimization imbalance caused by hard samples are proposed, demonstrating that HDF significantly improves both recognition accuracy and robustness.
- Cluster-Phys: Facial Clues Clustering Towards Efficient Remote Physiological Measurement29
A novel clustering-driven remote physiological measurement framework called Cluster-Phys is proposed, which employs a facial ROI prototypical clustering module to adaptively cluster the representative facial ROI features as facial prototypes and then update facial prototypes with highly semantic correlated base ROI features.
- 29
- 29
- Unified Static and Dynamic Network: Efficient Temporal Filtering for Video Grounding28
A novel residual structure (ResMLP) is devised to boost the global comprehensive interaction between the video segments and queries, achieving more effective semantic enhancement/supplement and the UniSDNet achieves SOTA performance on three widely used datasets for NLVG, as well as three datasets for SLVG.
- 28
- Early-Learning regularized Contrastive Learning for Cross-Modal Retrieval with Noisy Labels27
The proposed Early Learning regularized Contrastive Learning method for Cross Modal Retrieval with Noisy Labels (ELRCMR) is evaluated, and the result shows it could solve the cluster drift in conventional solutions and achieve promising performance on widely used benchmark datasets.
- Cross-Modality Retrieval by Joint Correlation Learning26
This article proposes a cross-modal learning model with joint correlative calculation learning that achieves comparable performance to the state-of-the-art on three benchmarks, i.e., Flickr8k, Flickr30k, and MS-COCO.
- LCSNet: End-to-end Lipreading with Channel-aware Feature Selection25
An end-to-end deep neural network-based lipreading model (LCSNet) that extracts the short-term spatio-temporal features and the motion trajectory features from the lip region in the video clips and achieves improvement of 52% and 47% compared to LipNet.
- Multimodal Counterfactual Learning Network for Multimedia-based Recommendation24
This work proposes a novel model called Multimodal Counterfactual Learning Network (MCLN), in which user-uninteracted items' multimodal content is additionally exploited to further purify the representation of user preference-relevant multimodAL content that better matches the user's interests, yielding state-of-the-art performance.
- Graph Pooling Inference Network for Text-based VQA24
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 and can effectively eliminate object redundancy and enhance the association of semantically continuous tokens.
- 22
- 21
- 21
- MAC 2024: Micro-Action Analysis Grand Challenge21
This grand challenge aims to foster innovative research in micro-action analysis and provide benchmark evaluations to advance the technology in the human-centric action understanding community.
- A sensing policy based on the statistical property of licensed channel in cognitive network21
This paper proposes a modified sensing policy based on the statistical property of licensed channels that can maintain sensing efficiency without adding constraints and also guarantee that more available licensed channels are available and the computational cost can be reduced.
- 20
- Integrated Agent-Based Modeling with GIS for Large Scale Emergency Simulation19
In the original version of this paper, the pages 618, 619, 622, 623 & 625 don't have the below additional information.
- Maskable Retentive Network for Video Moment Retrieval18
This paper introduces a new retention mechanism into the multimodal Transformer architecture, incorporating modality-specific attention modes into the multimodal Transformer architecture to address two key challenges in MR tasks: cross-modal guidance and video sequence modeling.
- Linguistics-Vision Monotonic Consistent Network for Sign Language Production16
A Transformer-based Linguistics-Vision Monotonic Consistent Network (LVMCN) is proposed for SLP, which constrains fine-grained cross-modal monotonic alignment and coarse-grained multimodal semantic consistency in language-visual cues through Cross-modal Semantic Aligner (CSA) and Multimodal Semantic Comparator (MSC).
- 16
- MHC-MAC: Cognitive MAC with asynchronous-assembly line mode for improving spectrum utilization and network capacity16
This poster presents a probabilistic procedure to establish an experimental procedure to count the number of particles in the response of the immune system to carbon dioxide.
- 15
- Pairwise VLAD Interaction Network for Video Question Answering15
A pairwise VLAD interaction mechanism is proposed to better exploit complementary information for each pair of modality descriptors, avoiding modeling uninformative individual relations, and exploring both inter- and intra-modality relations simultaneously.
- Micro-expression recognition with attention mechanism and region enhancement14
Attention mechanisms are introduced to enhance the representation performance of the model, which can improve the recognition accuracy and combine the facial RoI features with global facial features to recognize micro-expressions.
- 14
- Towards Energy-efficient Audio-visual Classification via Multimodal Interactive Spiking Neural Network13
A new Multimodal Interaction Spiking Network (MISNet) is presented, the first to successfully balance both accuracy and efficiency for the AVC task, and proposes a Multimodal Leaky Integrate-and-fire (MLIF) neuron, which coordinates and synchronizes the spike activations of audiovisual signals within a single neuron.
- 13
- Motion-compensated frame interpolation with weighted motion estimation and hierarchical vector refinement13
A novel MCFI algorithm based on weighted motion estimation (Weighted ME), motion vector (MV) segmentation, and hierarchical vector refinement that achieves a high image quality both subjectively and objectively is proposed.
- MA-Bench: Towards Fine-grained Micro-Action Understanding12
MA-Bench is presented, a benchmark comprising 1,000 videos and a three-tier evaluation architecture that progressively examines micro-action perception, relational comprehension, and interpretive reasoning that aims to establish a foundation benchmark for advancing MLLMs in understanding subtle micro-action and human-related behaviors.
- 12
- 12
- 12
- Scene-Text Grounding for Text-Based Video Question Answering12
The T2S-QA model is proposed that highlights a disentangled temporal-to-spatial contrastive learning strategy for weakly-supervised scene-text grounding and grounded TextVideoQA, thus decoupling QA from scene-text recognition and promoting research towards interpretable QA.
- Online Micro-gesture Recognition Using Data Augmentation and Spatial-Temporal Attention12
The latest solution developed by the HFUT-VUT team, HFUT-VUT, for the Micro-gesture Online Recognition track of the IJCAI 2025 MiGA Challenge is introduced, and the method ranked first in the Micro-gesture Online Recognition track.
- Unsupervised video summarization via clustering validity index12
This paper proposes an unsupervised video summarization method by motion-based frame selection and a novel clustering validity indexes to determine the optimal representatives of the original video.
- 12
- 11
- XInsight: Integrative Stage-Consistent Psychological Counseling Support Agents for Digital Well-Being11
Experiments show improved paradigm alignment, multi-therapy integration, interaction depth, and interpretability over existing multi-agent counseling systems, indicating that XInsight provides a practical blueprint for integrating counseling-inspired support agents into web applications for digital well-being.
- 11
- 11
- Continuous Sign Language Recognition Based on Pseudo-supervised Learning11
This work uses a 3D residual convolutional network pre-trained on the UCF101 dataset to extract visual features, and employs a sequence model with connectionist temporal classification (CTC) loss for learning the mapping between the visual features and sentence-level labels, which can be used to generate clip-level pseudo-labels.
- 11
- A BIT-PARALLEL ALGORITHM FOR SEQUENTIAL PATTERN MATCHING WITH WILDCARDS11
In BPBM, an extended bit-parallelism operation is used to accelerate the matching process and an effective transition window mechanism with two nondeterministic finite state automatons is adopted to drop the useless scan window.
- Towards Understanding Future: Consistency Guided Probabilistic Modeling for Action Anticipation10
A Consistency-guided Probabilistic Model (CPM), which focuses on learning the globally temporal probabilistic consistency to inhibit the unexpected temporal consistency, is proposed and achieves state-of-the-art performance on Epic-Kitchen100, Epic-Kitchen55, and EGTEA-GAZE.
- Visual-linguistic-stylistic Triple Reward for Cross-lingual Image Captioning10
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.
- Construction and Simulation of a Strategic HR Decision Model Based on Recurrent Neural Network10
RNN (Recurrent Neural Network) algorithm is used to conduct an in‐depth analysis of HR strategic decision‐making and an HR strategic decision model is constructed for simulation and shows that the established Bayesian model beats the platform’s built‐in intelligent method with a 78.2% win rate.
- Parametric and nonparametric residual vector quantization optimizations for ANN search10
Two optimized solutions based on residual vector quantization (RVQ) simulates restoring quantization error by multi-stage quantizers instead of decomposing it to further optimize codebook and space decomposition.
- 9
- Improved marching tetrahedra algorithm based on hierarchical signed distance field and multi-scale depth map fusion for 3D reconstruction9
Experimental results indicate that coarse search makes matching time shorter; using the H-SDF to fuse multi-scale depth maps, and isosurface extraction with improved marching tetrahedra algorithm can improve visual effect.
- 8
- Effectiveness of a Four-Stage Death Education Model Based on Constructivist Learning Theory for Trainee Nursing Students8
The four-step death education model based on constructivist learning theory significantly improved the death coping skills and attitudes of trainee nursing students and provides research materials for the future establishment of a systematic death education.
- Temporal Boundary Awareness Network for Repetitive Action Counting7
This work constructs a boundary-aware encoder equipped with a temporal pyramid structure to build multi-scale video features, capturing the period information of different lengths of repetitive actions to solve the period inconsistency problem.
- MAC 2025: The 2nd Micro-Action Analysis Grand Challenge7
An overview of the 2nd Micro-Action Analysis Grand Challenge is presented, including its dataset, evaluation protocol, results, and discussion, and potential future directions are outlined to guide ongoing developments in this area.
- DDFPN: Context enhanced network for object detection7
This work exploits an end-to-end Dilated and Deformable Feature Pyramid Network, namely DDFPN, to jointly extract spatial and semantic context and designs a Multi-scale DDC module to learn features for the various deformable objects.
- Co-occurrence pattern mining based on a biological approximation scoring matrix7
The co-fp-miner algorithm is an unified framework for both exact and approximate mining on multiple sequences and is more efficient on solutions, time and memory consumption than that of other peers.
- Complex-query web image search with concept-based relevance estimation7
A new image reranking scheme based on concept relevance estimation, which consists of Concept-Query and Concept-Image probabilistic models, that involves concepts by leveraging lexical and corpus-dependent knowledge with co-occurrence statistics of tags in the Flickr corpus.
- MAIL: mining sequential patterns with wildcards7
This paper designs an efficient mining algorithm MAIL, which mines four times more patterns than one of its peers and the time performance is six times faster on average than its another peer.
- 6
- SUEDE: Shared Unified Experts for Physical- Digital Face Attack Detection Enhancement6
Inspired by the efficacy of Mixture-of-Experts (MoE) in learning across diverse domains, SUEDE combines a shared expert to capture common features for both attack types and multiple routed experts (selectively activated) for specific attack types.
- 6
- Spatiotemporal contrastive modeling for video moment retrieval6
A new VMR framework, named spatiotemporal contrastive network (STCNet), is proposed, to learn discriminative boundary features of video grounding by contrast learning, and a boundary matching sampling module for dense negative sample sampling is proposed.
- Speedy and accurate image super‐resolution via deeply recursive CNN with skip connection and network in network6
A novel SISR method using a deeply recursive CNN with skip connections and a network in network structure is proposed, which significantly outperforms the previous methods by a large margin, while demanding far fewer parameters.
- Moderating the Generalization of Score-Based Generative Model5
The first Moderated Score-based Generative Model (MSGM) is proposed, which introduces a novel score adjustment strategy that redirects the score function away from undesirable data during the continuous-time stochastic differential equation process.
- Depth Matters: Spatial Proximity-Based Gaze Cone Generation for Gaze Following in Wild5
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, and employs the KL-divergence loss to super the grid-level regularized attention, providing comprehensive supervision for model optimization.
- Exploiting Diverse Feature for Multimodal Sentiment Analysis5
This paper proposes a series of feature extraction methods to build a robust representation and model ensemble in the MuSe 2023 Multimodal Sentiment Analysis Challenge and empirically evaluates the performance of the utilized method on the officially provided dataset.
- [Retracted] Analysis of Performance Evaluation Model for Higher Education Informatization by Means of PLS‐BP Model5
From the comparison results, the error between the PLS-BP evaluation model and traditional performance evaluation is relatively small, and the maximum relative error is 2.32%.
- Visual feature synthesis with semantic reconstructor for traditional and generalized zero‐shot object classification5
A novel named semantic consistent Wasserstein generative adversarial network (scWGAN) is proposed, which uses a semantic reconstructor to reconstruct semantic embeddings from generated visual features by incorporating a novel Semantic Consistent Loss noted L rec .
- 5
- MSPhys: multiscale fusing-based diffusion model for remote physiological measurement4
A MultiScale Fusing-based diffusion-based architecture to effectively capture the multiscale temporal contextual and periodic rPPG clues for precise rPPG estimation, and introduces a hierarchical loss function to constrain multiscale temporal consistency between the predicted and ground-truth rPPG signals, further improving estimation accuracy.
- 4
- Capacity-Constrained Contraflow Adaption for Lane Reconfiguration in Evacuation Planning4
Experimental results show that CC-Adap generates high-performing evacuation strategies and can be used to tackle large-scale evacuation planning.
- 4
- 3
- 3
- MAC 2026: Advancing Micro-Action Analysis Towards Fine-Grained Understanding3
A new task named fine-grained micro-action understanding is introduced, evaluated with the assistance of multimodal large language models, aiming to assess models'ability to capture fine-grained semantic cues and interpret subtle human micro-actions at a deeper level.
- 3
- 3
- Dual-path temporal map optimization for make-up temporal video grounding3
An effective proposal-based framework named Dual-Path Temporal Map Optimization Network is proposed to capture fine-grained multimodal semantic details of make-up activities to enhance the accuracy of video timestamp prediction.
- Semantic Enhanced Encoder-Decoder Network (SEN) for Video Captioning3
A semantic enhanced encoder-decoder network that implements a three path fusion strategy in the encoder side which combines complementary features and uses the idea of reinforcement learning to calculate rewards based on semantic designed computation.
- 3
- 3
- Performance Analysis of the Multiple Antenna Asynchronous Cognitive MAC Protocol in Cognitive Radio Network for IT Convergence3
A Multiple Antenna Asynchronous Cognitive MAC (MAAC-MAC) protocol is proposed through introducing a multiple antenna architecture and power control mechanism into the hardware-constrained cognitive MAC protocol with Asynchronous-Assembly Line Mode in cognitive radio network.
- 3
- 3
- Introduction to the Special Issue on Deep Learning for Robust Human Body Language Understanding2
This editorial introduces the Special Issue on Deep Learning for Robust Human Body Language Understanding, and highlights recent advancements in deep learning techniques that enhance the robustness, scalability, and applicability of human body language understanding in diverse scenarios, including healthcare, education, and barrier-free human-computer interaction systems.
- 2
- Alleviating Confirmation Bias in Learning with Noisy Labels via Two-Network Collaboration2
A novel method, TNCollab, which aims at alleviating confirmation bias in both sample selection and semi-supervised training stages via two-network collaboration, and introduces a class-adaptive threshold for sample selection to address the varying fitting difficulties across different classes.
- Multimodal Depression Estimation via Contrastive Modality Alignment and Fusion2
This work proposes a Contrastive Modality Alignment and Fusion (CMAF) framework, which outperforms advanced depression estimation methods and highlights the essence of effective alignment and fusion.
- Repetitive Action Counting with Feature Interaction Enhancement and Adaptive Gate Fusion2
This work proposes a feature interaction enhancement module designed to improve the learning of action-relevant features from both RGB and skeleton data, and introduces an adaptive local-global attention gating fusion mechanism to integrate enhanced processed RGB and skeleton features.
- Syntax-Controllable Video Captioning with Tree-Structural Syntax Augmentation2
A novel Tree-Structural Syntax Augmentation Network (TSAN) is proposed that consists of a tree-structural syntax encoder to capture syntactic parent-child relationships of example sentences and a syntax-customized caption generator to integrate syntax and semantic hints effectively.
- Evaluation of Tracheal Stenosis in Rabbits Using Multispectral Optoacoustic Tomography2
PAT and MSOT are potentially valuable non-invasive modality which are capable of evaluating tracheal structure and function in vivo, and both images and corresponding histology indicated the most severe degree of stenosis in group C.
- 2
- 2
- 2
- 2
- Implement of the 3D Virtual Mine System Based on OpenGL2
The method of constructing 3D virtual mine system based on OpenGL with great significance in doing geological analysis and improving mine production is advanced and the functions and the overall structure frame are introduced.
- 1
- 1
- 1
- Face-Guided Sentiment Boundary Enhancement for Weakly-Supervised Temporal Sentiment Localization1
The Face-guided Sentiment Boundary Enhancement Network (FSENet), a unified framework that leverages fine-grained facial features to guide sentiment localization, is proposed, achieving state-of-the-art performance under full supervision, video-level, and point-level weak supervision.
- 1
- 1
- 1
- 1
- MER-Bench: A Comprehensive Benchmark for Multimodal Meme Reappraisal1
A novel multimodal generation task that aims to transform negatively framed memes into constructive ones while preserving their underlying scenario, entities, and structural layout, and a structured evaluation framework based on a multimodal large language model (MLLM)-as-a-Judge paradigm are introduced.
- Bidirectional Counterfactual Distillation for Review-Based Recommendation1
This work proposes a novel Bidirectional Counterfactual Distillation (BiCoD) framework for review-based recommendation, and introduces a stage-aware bidirectional distillation strategy to enhance the distinctiveness of behavioral features, facilitating the effective learning of behavior-specific preferences.
- 1
- Every Subtlety Counts: Fine-grained Person Independence Micro-Action Recognition via Distributionally Robust Optimization1
The Person Independence Universal Micro-action recognition Framework is proposed, which integrates Distributionally Robust Optimization principles to learn person-agnostic representations and outperforms existing methods in both accuracy and robustness, achieving stable generalization under fine-grained conditions.
- Reproducibility Companion Paper: Maskable Retentive Network for Video Moment Retrieval1
This paper presents artifacts that enable the reproduction of the results from earlier work, including details on dataset preparation, source code, and the experimental environment, and demonstrates the reproducibility of the results.
- 1
- 1
- Distilling Textual Priors From LLM to Efficient Image Fusion1
This work proposes a novel framework for distilling large model priors, eliminating the need for text guidance during inference while dramatically reducing model size, and introduces a spatial-channel cross-fusion module to enhance the model's ability to leverage textual priors across both spatial and channel dimensions.
- EmoSEM: Segment and Explain Emotion Stimuli in Visual Art1
The Emotion stimuli Segmentation and Explanation Model (EmoSEM) model is proposed, which realizes end-to-end modeling from low-level pixel features to high-level emotion interpretation, delivering the first interpretable fine-grained framework for visual emotion analysis.
- DFGAP: Towards Depth-Free Cross-Category GAParts Perception via Uncertainty-Quantified Modeling1
This paper proposes DFGAP, a novel depth-free framework for RGB-based GAParts segmentation and pose estimation, and independently model the ill-pose problems from the absence of depth for GAPart segmentation and pose estimation, by clearly quantifying the pixel-wise segmentation probability and relative depth.
- AMTN: Attention-Enhanced Multimodal Temporal Network for Humor Detection1
The Attention-Enhanced Multimodal Temporal Network (AMTN) is introduced to address the MuSe 2024 Cross-Cultural Humor Detection Sub-Challenge (MuSe-Humor), which highlights the task of humor detection within a cross-cultural context, leveraging multimodal information for its accomplishment.
- 1
- Detection of defects on the end face of small motor bearings based on machine vision1
The use of machine vision and image processing techniques to enhance the speed, efficiency, and reliability of detection of bearing defect detection and improve detection efficiency through the advancement of deep learning-based target detection algorithms is proposed.
- DQKNet: Deep Quasiconformal Kernel Network Learning for Image Classification1
The experimental results demonstrate that the improvement in the deep kernel mapping network from the perspectives of accounting children, mapping network nodes, and network structure can effectively enhance the feature extraction and classification performance of the data.
- 1
- 1
- 1
- 1
- 1
- Study on the Reform of Computer Practice Teaching1
The paper proposed the reform measures of computer practice reaching basing on the actual need in teaching work that will raise the students’ creative and practice abilities, promote the students adapt to the social requests, and help the students work and develop better.
- 1
- A Decision-Making Model for Distribution Center Location during Earthquake Responses1
This paper investigates the factors impacting the location of distribution center during earthquake responses, and proposes a decision-making model based on improved AHP framework for this issue that was successfully applied in the simulation case of Yushu earthquake responses in 2010.
- 1
- 1
- –
- –
- –
- –
- LinProVSR: Linguistics-Knowledge Guided Progressive Disambiguation Network for Visual Speech Recognition–
The Ambiguous Word Error Rate (AWER) metric is pioneer specifically for evaluating recognition of phonetically ambiguous text, and the effectiveness of the proposed LinProVSR framework is verified on multiple public datasets, achieving a significant breakthrough especially in distinguishing visually similar phonemes.
- –
- –
- SIAM: Towards Generalizable Articulated Object Modeling via Single Robot-Object Interaction–
SIAM achieves accurate, prior-free modeling with significantly reduced interaction cost, and reconstructs a high-quality, topologically watertight mesh by learning 3D Gaussian Primitives from multi-view RGB-D observations under deformation.
- SIGaze: Toward Pixel-Level Single-Instance Gaze Object Prediction–
The proposed SIGaze method significantly outperforms existing methods in the single-instance gaze object prediction task, and extensive experiments on the GOO dataset demonstrate that this method significantly outperforms existing methods in the single-instance gaze object prediction task.
- Learning Speaker-Invariant Visual Features for Lipreading–
SIFLip is a speaker-invariant visual feature learning framework that disentangles speaker-specific attributes using two complementary disentanglement modules (Implicit Disentanglement and Explicit Disentanglement) to improve generalization.
- KineDiff3D: Kinematic-Aware Diffusion for Category-Level Articulated Object Shape Reconstruction and Generation–
This work proposes KineDiff3D: Kinematic-Aware Diffusion for Category-Level Articulated Object Shape Reconstruction and Generation, a unified framework for reconstructing diverse articulated instances and pose estimation from single view input.
- –
- –
- Research on the object detection in complex scenes using diagonal distance–
A diagonal distance loss function, , and a combined diagonal distance loss function, based IoU, using the distance between diagonals as a penalty are proposed, overcoming the degradation phenomenon of prior functions based on traditional IoU.
- –
- –
- –
- –
- –
- –
- –
- Distilling Dynamic Spatial Relation Network for Human Pose Estimation–
A Distilling Dynamic Spatial Relation network (DDSR), which builds pose-based graph representation by exploiting the feature of spatial relation from the location distribution of joints by using a dynamic message propagation mechanism to update the spatial relation on edges.
- –
- –
- –
- Piecewise-recursive Convolutional Network for Fast and Accurate Face Image Super-resolution–
Experimental results show that the proposed algorithm outperforms the state-of-the-art methods, while achieving faster and more efficient computation.
- –
- Implementation and management of clinical research training–
The management of DCR online course serves as a great exploration on training junior medical professionals and international research collaboration and junior talents training.
- –
- –
- Research on Software Architecture Base on Data Snap Technology–
Using the software architecture, system can improve the software reusability, meanwhile, can solve the system load overweight cause by data access operations.
- –
- –
- –
- –
- The solution to knapsack problem with genetic greedy algorithm–
The method is able to overcome premature convergence of genetic algorithms and show the better performance of the improved algorithm for solving knapsack problem.
- Applicablity and Case-study of Gray Correlation Analysis in High-tech Enterprises–
This article’s innovation comes from refer the gray relational analysis to high-tech enterprise performance evaluation, and provide new ideas and methods for the high-Tech enterprises’ performance evaluation.
- –
- Visual Literacy Training in Vocational Education University–
The method of using knowledge visualization tools to provide effective visual literacy training for the students in vocational education universities is explored and the results have been satisfactory and the authors are working on further improving it.
- –
- –
- –
- –
- A Classifying and Searching Method for Efficient–
A searching method for eficient subset for portfolio selection is obtained, which is proved valid, based on classifying efficient srbset for portfolioselection.
- –
- –
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.