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
- Pong Chi YuenSuggested from co-authorship
- Rama ChellappaSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works97 from public data
- The Visual Object Tracking VOT2016 Challenge Results776
The Visual Object Tracking challenge VOT2016 goes beyond its predecessors by introducing a new semi-automatic ground truth bounding box annotation methodology and extending the evaluation system with the no-reset experiment.
- 435
- Hierarchical Discriminative Learning for Visible Thermal Person Re-Identification422
An improved two-stream CNN network is presented to learn the multi-modality sharable feature representations and identity loss and contrastive loss are integrated to enhance the discriminability and modality-invariance with partially shared layer parameters.
- Visible Thermal Person Re-Identification via Dual-Constrained Top-Ranking413
A dual-path network with a novel bi-directional dual-constrained top-ranking loss to learn discriminative feature representations and identity loss is further incorporated to model the identity-specific information to handle large intra-class variations.
- 331
- 235
- 217
- 208
- Regularized Fine-Grained Meta Face Anti-Spoofing195
This work casts face anti-spoofing as a domain generalization (DG) problem, and attempts to address this problem by developing a new meta-learning framework called Regularized Fine-grained Meta-learning, which incorporates the domain knowledge of faceAnti-sp spoofing as the regularization so that meta- learning is conducted in the feature space regularized by the supervision of domain knowledge.
- 156
- 130
- The Ninth Visual Object Tracking VOT2021 Challenge Results127
The Visual Object Tracking challenge VOT2021 is the ninth annual tracker benchmarking activity organized by the VOT initiative; results of 71 trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in recent years.
- Learning Modality-Consistency Feature Templates: A Robust RGB-Infrared Tracking System123
A new tracking system, which aims to combine the information from RGB and infrared modalities for object tracking, based on the proposed machine learning model that can alleviate the modality discrepancy issue under the proposed modality consistency constraint.
- Remote Photoplethysmography Correspondence Feature for 3D Mask Face Presentation Attack Detection123
A new liveness feature, called rPPG correspondence feature (CFrPPG) is proposed to precisely identify the heartbeat vestige from the observed noisy rP PG signals to overcome the global interferences.
- Modality-correlation-aware sparse representation for RGB-infrared object tracking119
This paper presents a feature representation and fusion model to combine the feature representation of the object in RGB and infrared modalities for object tracking and demonstrates the effectiveness of the proposed method.
- CricaVPR: Cross-Image Correlation-Aware Representation Learning for Visual Place Recognition115
A robust global representation method with cross-image correlation awareness for VPR, named CricaVPR, which uses the attention mechanism to correlate multiple images within a batch and introduces the multi-scale local information to further enhance the cross-image correlation-aware representation.
- Modality-aware Collaborative Learning for Visible Thermal Person Re-Identification114
A novel modality-aware collaborative (MAC) learning method on top of a two-stream network for VT-ReID, which handles the modalities-discrepancy in both feature level and classifier level, and introduces a collaborative learning scheme, which regularizes themodality-sharable and modality -specific identity classifiers by utilizing the relationship between different classifiers.
- Robust Anchor Embedding for Unsupervised Video Person re-IDentification in the Wild111
This paper addresses the scalability and robustness issues of estimating labels from imbalanced unlabeled data for unsupervised video-based person re-identification (re-ID) and proposes a novel Robust AnChor Embedding (RACE) framework via deep feature representation learning for large-scale un Supervised video re-ID.
- 103
- Towards Seamless Adaptation of Pre-trained Models for Visual Place Recognition100
A hybrid adaptation method to achieve both global and local adaptation efficiently, in which only lightweight adapters are tuned without adjusting the pre-trained model, and a mutual nearest neighbor local feature loss, which ensures proper dense local features are produced for local matching and avoids time-consuming spatial verification in re-ranking.
- 95
- 92
- 86
- 85
- 80
- Robust Collaborative Discriminative Learning for RGB-Infrared Tracking77
A novel and optimal discriminative learning framework for multi-modality tracking that is able to jointly eliminate outlier samples caused by large variations and learn discriminability-consistent features from heterogeneous modalities, and collaboratively perform modality reliability measurement and target-background separation.
- Robust joint discriminative feature learning for visual tracking76
This paper proposes an unified feature learning framework which simultaneously exploits both the representation capability and the discriminability of multiple features for visual tracking and performs favourably against other state-of-the-art trackers.
- Toward Visual Grounding: A Survey72
This survey examines the developmental history of visual grounding and provides an overview of essential background knowledge, including fundamental concepts and evaluation metrics, and encompasses the representative work in each subtopic over the past decade.
- Robust MIL-Based Feature Template Learning for Object Tracking53
This paper proposes a novel and optimal model learning (updating) scheme which aims to simultaneously eliminate the negative effects from these two issues mentioned above in a unified robust feature template learning framework.
- Continuous Prediction of Lower-Limb Kinematics From Multi-Modal Biomedical Signals52
This is the first study of continuously predicting lower-limb kinematics in an ahead-of-time manner based on the electromechanical delay (EMD), and the time boundary of compensating the hysteretic response is revealed.
- Point-to-Set Distance Metric Learning on Deep Representations for Visual Tracking52
This paper proposes to integrate the point-to-set/ image- to-imageSet distance metric learning (DML) into visual tracking tasks and take full advantage of all the training samples when determining the best target candidate.
- 47
- Multimodal PEAR Chain-of-Thought Reasoning for Multimodal Sentiment Analysis45
Experimental results on two multimodal sentiment analysis benchmark datasets show that high-level reasoning information can help learn discriminative text representation, and cross-modal complementary information can avoid misleading by unreasonable steps in the CoT.
- Driver distraction detection using capsule network45
A CapsNet-based approach for detecting the distracted driver which is a novel approach that performs well on the real-world environment inputs and gets high scores for all the most commonly used metrics for classification.
- Enhancing LLM Reasoning with Iterative DPO: A Comprehensive Empirical Investigation40
It is demonstrated that a single round of DPO with coarse filtering significantly enhances mathematical reasoning performance, particularly for strong base model, and an iterative enhancement framework for both the generator and the reward model is designed, enabling their mutual improvement through online interaction across multiple rounds of DPO.
- 40
- AlignMamba: Enhancing Multimodal Mamba with Local and Global Cross-Modal Alignment39
Align-Mamba is proposed, an efficient and effective method for multimodal fusion that introduces a local cross-modal alignment module that explicitly learns token-level correspondences between different modalities and proposes a global cross-modal alignment loss based on Maximum Mean Discrepancy to implicitly enforce the consistency between different modal distributions.
- 39
- OV-DINO: Unified Open-Vocabulary Detection with Language-Aware Selective Fusion38
A novel unified open-vocabulary detection method called OV-DINO is proposed, which is pre-trained on diverse large-scale datasets with language-aware selective fusion in a unified framework, and a Unified Data Integration pipeline is introduced to enable end-to-end training and eliminate noise from pseudo-label generation by unifying different data sources into detection-centric data format.
- Cohesive Multi-Modality Feature Learning and Fusion for COVID-19 Patient Severity Prediction38
A multi-modality feature learning and fusion model for end-to-end covid patient severity prediction using the blood test supported electronic medical record (EMR) and chest computerized tomography (CT) scan images is proposed.
- Deep Refinement: capsule network with attention mechanism-based system for text classification37
This paper proposes a pipeline called Deep Refinement which utilizes some of the state-of-the-art methods for information retrieval from highly sparse data such as capsule network and attention mechanism to classify text primarily into two categories, namely sincere and insincere.
- 32
- Online Non-Negative Multi-Modality Feature Template Learning for RGB-Assisted Infrared Tracking28
An infrared tracking framework in which information from RGB-modality will be exploited to assist the infrared object tracking, and an online non-negative feature template learning model is designed in order to deal with the contaminated features caused by large appearance variations.
- Double-Domain Imaging and Adaption for Person Re-Identification28
A novel framework—double-domain translation generative adversarial network (DTGAN) that can train images between two domains and generalize the model trained on one domain well to another domain and moderating the influence of noise is proposed.
- 27
- 26
- Deep Homography Estimation for Visual Place Recognition25
A transformer-based deep homography estimation (DHE) network that takes the dense feature map extracted by a backbone network as input and fits homography for fast and learnable geometric verification is proposed and can outperform several state-of-the-art methods.
- SuperVLAD: Compact and Robust Image Descriptors for Visual Place Recognition24
Experimental results suggest that, when paired with a transformer-based backbone, the SuperVLAD shows better domain generalization performance than NetVLAD with significantly fewer parameters, and surpasses state-of-the-art methods with lower feature dimensions on several benchmark datasets.
- Feature Constrained by Pixel22
A new model based on Generative Adversarial Network is proposed, named HierarchicalAdversarial Deep Network (HADN), which jointly optimizes the feature-level and pixel-level adversarial adaptation within a hierarchical network structure.
- SelaVPR++: Towards Seamless Adaptation of Foundation Models for Efficient Place Recognition20
An extension of the SelaVPR, called SelaVPR++, is proposed that is highly efficient in training time, GPU memory usage, and retrieval latency, as well as outperforms the state-of-the-art methods by a large margin.
- 19
- Strip-MLP: Efficient Token Interaction for Vision MLP19
Strip-MLP significantly improves the performance of MLP-based models on small datasets and obtains comparable or even better results on ImageNet.
- Robust Multi-modality Anchor Graph-based Label Prediction for RGB-Infrared Tracking19
A robust multi-modality anchor graph-based label prediction model is developed, which is able to construct a scalable graph representation of the relationship of the samples based on local anchor approximation and defuse a limited amount of known labels to large amount of unlabeled sample efficiently based on transductive learning strategy.
- 18
- Fusion of iris and sclera using phase intensive rubbersheet mutual exclusion for periocular recognition18
A new multi-biometric fusion method called Phase Intensive Mutual Exclusive Distribution (PI-MED) method by combining periocular features for identity verification by reducing the matching fusion time and overhead during human recognition in biometrics is presented.
- VER-Bench: Evaluating MLLMs on Reasoning with Fine-Grained Visual Evidence16
The VER-Bench is introduced, a novel framework to evaluate MLLMs' ability to identify fine-grained visual clues and integrate these clues with world knowledge for complex reasoning, highlighting the need to enhance models' capabilities in fine-grained visual evidence extraction, integration, and reasoning for genuine visual understanding and human-like analysis.
- High-Resolution Image Harmonization with Adaptive-Interval Color Transformation16
An Adaptive-Interval Color Transformation method (AICT), which predicts pixel-wise color transformations and adaptively adjusts the sampling interval to model local non-linearities of the color transformation at high resolution.
- 16
- Detecting and Tracking of Multiple Mice Using Part Proposal Networks14
This article proposes an efficient and robust deep-learning-based mouse part detection scheme and proposes a novel Bayesian-inference integer linear programming model that jointly assigns the part candidates to individual targets with necessary geometric constraints while establishing pair-wise association between the detected parts.
- 14
- Limb-Aware Virtual Try-On Network With Progressive Clothing Warping13
Lb-aware Virtual Try-on Network named PL-VTON is proposed, which performs fine-grained clothing warping progressively and generates high-quality try-on results with realistic limb details and introduces Limb-aware Texture Fusion that focuses on generating realistic details in limb regions.
- Spatial-temporal Regularized Multi-modality Correlation Filters for Tracking with Re-detection12
An adaptive spatial-temporal regularization is imposed into the correlation filter framework in which the spatial regularization can help to suppress effect from the cluttered background while the temporal regularization enables the adaptive incorporation of historical appearance cues to deal with appearance changes.
- LRDNN: Local-refining based Deep Neural Network for Person Re-Identification with Attribute Discerning12
A Local-refining based Deep Neural Network (LRDNN) is proposed to aggregate pose estimation and attribute recognition to improve the re-ID performance and diminish the negative affect from unstable pose estimation.
- UP-Person: Unified Parameter-Efficient Transfer Learning for Text-Based Person Retrieval11
This paper proposes a novel Unified Parameter-Efficient Transfer Learning (PETL) method for Text-based Person Retrieval (UP-Person) to thoroughly transfer the multi-modal knowledge from CLIP, which simultaneously integrates three lightweight PETL components including Prefix, LoRA and Adapter.
- Towards Adaptable Graph Representation Learning: An Adaptive Multi-Graph Contrastive Transformer10
An adaptive multi-graph contrastive transformer (AMGCT) is introduced for general spatial-temporal graph representation learning and a coupled graph transformer (CGT) is designed to consider spatial relationships at each stage of temporal modeling, explore complementary information between spatial and temporal domains, and obtain more compact spatial-Temporal representations.
- 8
- EMMA: Empowering Multi-modal Mamba with Structural and Hierarchical Alignment7
This work proposes Empowering Multi-modal Mamba with Structural and Hierarchical Alignment (EMMA), which enables the MLLM to extract fine-grained visual information and shows lower degrees of hallucination and enhanced sensitivity to visual details, which manifests in superior performance across diverse multi-modal benchmarks.
- Calibration for Long-tailed Scene Graph Generation6
This work attempts to highlight and remedy miscalibration in current scene graph generation (SGG) models, which has been overlooked by previous works, and proposes the Compositional Optimization Calibration (COC) method, which performs best on model calibration compared to different types of calibration methods.
- ClickTrack: Towards real-time interactive single object tracking6
A new paradigm for single object tracking algorithms, ClickTrack, a new paradigm using clicking interaction for real-time scenarios, and the Guided Click Refiner (GCR), which accepts point and optional textual information as inputs, transforming the point into the bounding box expected by the operator.
- Cross-DINO: Cross the Deep MLP and Transformer for Small Object Detection5
Extensive experimental results on COCO, WiderPerson, VisDrone, AI-TOD, and SODA-D datasets demonstrate that Cross-DINO efficiently improves the performance of DETR-like models on SOD.
- 5
- 5
- Open-Det: An Efficient Learning Framework for Open-Ended Detection5
Open-Det accelerates model training in both the bounding box and object name generation process by reconstructing the Object Detector and the Object Name Generator, and proposes a Vision-Language Aligner with V-to-L and L-to-V alignment mechanisms, enabling accurate object name generation for the LLM.
- 5
- Adversarial auto‐encoder for unsupervised deep domain adaptation5
This study proposes a novel adversarial auto-encoder for unsupervised deep domain adaptation that incorporates the auto- Encoder with the adversarial learning so that the domain similarity and reconstruction information from the decoder can be exploited to facilitate the adversaria domain adaptation in the encoder.
- 4
- 4
- K-Space Bispectrum Steganography for Robust Unlearnable Data3
KBS (K-Space Bispectrum Steganography), which embeds class-specific information into the magnitude and phase components of the Fourier domain while preserving visual fidelity under reconstruction constraints, is proposed, which outperforms state-of-the-art methods.
- 3
- 3
- PaperAudit-Bench: Benchmarking Error Detection in Research Papers for Critical Automated Peer Review2
This paper introduces PaperAudit-Bench, an automated review framework that integrates structured error detection with evidence-aware review generation to support critical assessment, and shows that the dataset supports training lightweight LLM detectors via SFT and RL, enabling effective error detection at reduced computational cost.
- 2
- AlignMamba-2: Enhancing multimodal fusion and sentiment analysis with modality-aware Mamba1
AlignMamba-2 is proposed, an effective and efficient framework for multimodal fusion and sentiment analysis that introduces a dual alignment strategy that regularizes the model using both Optimal Transport distance and Maximum Mean Discrepancy, promoting geometric and statistical consistency between modalities without incurring any inference-time overhead.
- 1
- 1
- DS-Det: Single-Query Paradigm and Attention Disentangled Learning for Flexible Object Detection1
DS-Det is proposed, a more efficient detector capable of detecting a flexible number of objects in images and reformulate and introduce a new unified Single-Query paradigm for decoder modeling, transforming the fixed-query into flexible.
- Generic Scene Graph Generation Model with Hierarchical Prompt Learning1
GSGG (Generic Scene Graph Generation), a novel, efficient, and flexible SGG model that combines generalized modules to construct top-performance and high-efficiency SGG model and employs a prompt learning-based relation decoder with a novel Hierarchical Prompt (HP) learning method to mitigate biased optimization.
- 1
- 1
- –
- –
- –
- –
- –
- –
- –
- Multi-cue visual tracking: feature learning and fusion–
This thesis proposes a novel robust joint discriminative feature learning framework which is capable of simultaneously and optimally removing corrupted features and learning reliable classifiers with more multiple sparse representation model for multi-cue visual tracking which jointly exploits the underlying commonalities and diversities of different visual cues by decomposing multiple sparsity patterns.
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 2026-10-15. 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.