Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks150 from public data
- 183
- 159
- 118
- 112
- 104
- 95
- Learning Domain Invariant Prompt for Vision-Language Models93
This study introduces an innovative prompt learning paradigm, called MetaPrompt, aiming to directly learn domain invariant prompt in few-shot scenarios and introduces a novel asymmetric contrastive learning paradigm, where representations from the pre-trained encoder assume supervision to regularize prompts from the prompted encoder.
- 67
- Scene Text Image Super-Resolution via Parallelly Contextual Attention Network66
A Parallelly Contextual Attention Network (PCAN), which effectively learns sequence-dependent features and focuses more on high-frequency information of the reconstruction in text images, and outperforms eleven state-of-the-art (SOTA) SR methods in terms of boosting text recognition performance.
- A novel statistical image thresholding method64
A novel criterion combining between-class variance and minimizing class variance is proposed to eliminate the described limitation for classic statistical approaches and improve segmentation performance.
- Similarity learning with joint transfer constraints for person re-identification59
The issue of person re-identification under inconsistent data distributions can be transformed into a consistent iterative convex optimization problem, and solved via the inexact augmented Lagrange multiplier (IALM) algorithm.
- Multilevel triplet deep learning model for person re-identification57
This paper improved the performance of deep CNN network with the proposed Multilevel feature extraction strategy and built a novelMultilevel triplet deep learning model corresponding to the method.
- 56
- 53
- 49
- Maximum decision entropy-based attribute reduction in decision-theoretic rough set model47
A novel monotonic uncertainty measure is introduced for attribute reduction in the decision-theoretic rough set model and a heuristic attribute reduction algorithm based on the maximum decision entropy is developed, which maximizes the relevance of the reduct to the class attribute and also minimizes the redundancy of the condition attributes within the reductor.
- 46
- 45
- 44
- 42
- 34
- 32
- 31
- Multi-scale gist feature manifold for building recognition31
The results show that the proposed model is superior to other models in practice of building recognition and can handle the building recognition problem caused by rotations, variant lighting conditions and occlusions very well.
- Diverse Person: Customize Your Own Dataset for Text-Based Person Search30
This paper proposes to leverage available images of clothing and accessories as reference attribute images to edit the original dataset images through diffusion models, and employs a Large Language Model (LLM) to produce annotations that are both high in quality and stylistically consistent with those found in real-world datasets.
- Invisible Backdoor Attack With Dynamic Triggers Against Person Re-Identification30
A novel backdoor attack on deep ReID under a new all-to-unknown scenario, called Dynamic Triggers Invisible Backdoor Attack (DT-IBA), which instead of learning fixed triggers for the target classes from the training set, can dynamically generate new triggers for any unknown identities.
- 30
- 30
- Maximal granularity structure and generalized multi-view discriminant analysis for person re-identification29
The proposed descriptor of MGSD captures rich local structural information from overlapping macro-pixels in an image, analyzes the horizontal occurrence of multi-granularity and maximizes the occurrence to extract a robust representation for viewpoint changes, and can obtain rich person appearance whilst being robust against different condition changes.
- Two-dimensional color uncorrelated discriminant analysis for face recognition29
Experimental results show that the proposed algorithm performs better than other color face recognition methods and the two-dimensional color uncorrelated discriminant features are more effective for low-resolution image compared with conventional gray-based features.
- 28
- Uncertainty-optimized deep learning model for small-scale person re-identification27
This study considers the uncertainty of pedestrian representation for small-scale person re-identification and designs an improved Monte Carlo strategy that considers both the average distance and shortest distance for matching and ranking.
- Multiple metric learning based on bar-shape descriptor for person re-identification26
This paper proposes a new Multiple Bar-shape Descriptor that can take full account of the spatial correlation between the center points and their adjacent points on different directions and provides an effective multiple metric learning method fusing two-channel bar-shape structural features via the optimal similarity pairwise measure obtained by a dissimilarity matrix.
- FaceShot: Bring Any Character into Life25
FaceShot is a novel training-free portrait animation framework designed to bring any character into life from any driven video without fine-tuning or retraining, and is compatible with any landmark-driven animation model, significantly improving overall performance.
- 25
- Cross-Modal Distillation for Speaker Recognition23
This work introduces a multimodal learning framework, VGSR (Vision-Guided Speaker Recognition), and proposes a MKD (Margin-based Knowledge Distillation) strategy for cross-modality distillation by introducing a loose constrain to align the teacher and student, greatly reducing overfitting.
- Improved adaptive image retrieval with the use of shadowed sets23
A shadowed-set-based image retrieval algorithm is proposed, and techniques of an automatic selection of two threshold parameters by combining saliency detection and edge detection, which automatically determine shadowed regions are developed.
- Feature extraction using adaptive slow feature discriminant analysis23
A new feature extraction method called adaptive slow feature discriminant analysis (ASFDA), which provides an adaptive parameter to balance between-class temporal variation and within- class temporal variation to obtain an optimal discriminant subspace.
- Multiscale principal contour direction for varying lighting face recognition22
MPCD can capture ‘true’ neighbourhood contour direction information and is robust to different illumination conditions and is constructed for extracting illumination invariant by applying the nonsubsampled contourlet transform (NSCT) to multiscale contour analysis.
- 19
- 19
- 17
- 17
- Fuzzy local maximal marginal embedding for feature extraction17
An improved algorithm called fuzzy local maximal marginal embedding (FLMME) for linear dimensionality reduction is presented, which significantly differing from the existing graph-based algorithms is that two novel fuzzy gradual graphs are constructed in FLMME.
- Does Video-Text Pretraining Help Open-Vocabulary Online Action Detection?16
This research establishes a robust base-line for zero-shot transfer in online action detection, enabling scalable solutions for open-world temporal understanding and proposes OV-OAD, a zero-shot online action detector that leverages vision-language models and learns solely from text supervision.
- Uncorrelated slow feature discriminant analysis using globality preserving projections for feature extraction16
A novel method called uncorrelated slow feature discriminant analysis using globality preserving projections (USFDA-GP) is proposed for feature extraction and recognition and provides an analytical solution by simple eigenvalue decomposition to the optimal model instead of previous iterative method.
- 15
- Optimized Automatic Seeded Region Growing Algorithm with Application to ROI Extraction15
The proposed Optimized Automatic Seeded Region Growing (OASRG) algorithm uses the affinity propagation (AP) clustering algorithm to extract the seeds automatically, and optimizes the traditional region growing algorithm by regrowing strategy to obtain the regions of interest where target objects are contained.
- 15
- Kernelized random KISS metric learning for person re-identification14
Experimental results on three challenging person re-identification datasets demonstrate that the KRKISS method significantly improves the KISS method and achieves better performance than most existing metric learning approaches.
- Fusion of multiple channel features for person re-identification14
An effective feature representation called fusion of multiple channel features (FMCF) is proposed which captures different low-level features from multiple channels of HSV color space, considering the characteristics of different color channels and fusing color, texture and correlation of spatial structure.
- CharacterShot: Controllable and Consistent 4D Character Animation13
A controllable and consistent 4D character animation framework that enables any individual designer to create dynamic 3D characters from a single reference character image and a 2D pose sequence, and demonstrates that this approach outperforms current state-of-the-art methods.
- 13
- 13
- Fetch and Forge: Efficient Dataset Condensation for Object Detection11
This work proposes DCOD, the first dataset condensation framework for object detection, which operates in two stages: Fetch and Forge, initially storing key localization and classification information into model parameters, and then reconstructing synthetic images via model inversion.
- Towards Privacy-Preserving Person Re-identification via Person Identify Shift11
A novel de-identification method designed explicitly for person ReID, named Person Identify Shift (PIS), which removes the absolute identity in a pedestrian image while preserving the identity relationship between image pairs by exploiting the interpolation property of variational auto-encoder.
- Vehicle and wheel detection: a novel SSD-based approach and associated large-scale benchmark dataset11
A novel optimized SSD algorithm with multi-concatenation modules, aiming to improve the performance of small object detection, and experimental results show that the proposed method results in better detection performance on small objects.
- QRKISS: A Two-Stage Metric Learning via QR-Decomposition and KISS for Person Re-Identification11
This paper proposes a novel two-stage metric learning via QR-Decomposition and KISS, named QRKISS, which is better than other KISS-based metric learning methods and achieves state-of-the-art performance.
- 9
- A Neural Representation Framework with LLM-Driven Spatial Reasoning for Open-Vocabulary 3D Visual Grounding8
This work proposes SpatialReasoner, a novel neural representation-based framework with large language model (LLM)-driven spatial reasoning that constructs a visual properties-enhanced hierarchical feature field for open-vocabulary 3D visual grounding.
- Fisher Difference Discriminant Analysis: Determining the Effective Discriminant Subspace Dimensions for Face Recognition8
Comprehensive comparison and extensive experiments show that FDDA is superior to some state-of-the-art techniques in face recognition and robust to noise and invariant to rotations, rescalings and translations of the data.
- 7
- 7
- 7
- 7
- FLAG: feature learning with additional guidance for person search7
A novel method called FLAG is presented to learn a better feature representation for person search and partition pooling and cross-level feature hybridization are proposed to guide the model to learn more discriminative person features.
- A Novel Hard Mining Center-Triplet Loss for Person Re-identification7
The results on two large-scale datasets Market1501 and DukeMTMC-reID show the robustness and efficiency of the method in making optimizations of these problems simultaneously and learning robust feature representation, which demonstrate that the method outperforms most of existing loss function and achieves better performance for person re-identification.
- 7
- Automatic Image Annotation Based on Multi-Auxiliary Information7
An automatic image annotation framework based on multi-auxiliary information which aims at improving the annotation performance is introduced and the tag refinement process is formalized as an optimization problem by adjusting confidence score set by the initial labeling model.
- Graph embedding discriminant analysis for face recognition7
The experimental results show that GEDA consistently outperforms LDA and MFA when the training sample size per class is small, and is applied to face recognition and is examined on the Yale, ORL and AR face image databases.
- 6
- Re-ID-leak: Membership Inference Attacks Against Person Re-identification6
A novel MI attack method based on the distribution of inter-sample similarity, which involves sampling a set of anchor images to represent the similarity distribution that is conditioned on a target image, is proposed.
- 6
- Robust Color Invariant Model for Person Re-Identification6
A novel feature representation called Color Invariant Feature (CIF) is proposed, it is robust to illumination and viewpoint changes and outperforms the state of the art in person re-identification.
- Maximal local interclass embedding with application to face recognition6
A new dimensionality reduction approach called maximal local interclass embedding (MLIE) is developed, which can be viewed as a linear approach of a multimanifolds-based learning framework, in which the information of neighborhood is integrated with the local inter class relationships.
- 6
- 6
- 5
- Fusion4DAL: Offline Multi-modal 3D Object Detection for 4D Auto-labeling5
This work proposes a simple yet effective offline multi-modal 3D object detection method, named Fusion4DAL, for 4D auto-labeling with long-term multi-modal sequences, and proposes a multi-modal mixed feature fusion module (MMFF), to address the sparsity of points within objects.
- PatSTEG: Modeling Formation Dynamics of Patent Citation Networks via The Semantic-Topological Evolutionary Graph5
A joint semantic-topological evolutionary graph learning approach (PatSTEG) is proposed to model the formation dynamics of patent citation networks and proves the superiority of PatSTEG over other state-of-the-art methods.
- 5
- 5
- Integrating Disparity Confidence Estimation Into Relative Depth Prior-Guided Unsupervised Stereo Matching4
This work proposes a novel unsupervised learning framework to address stereo matching ambiguities, which comprises a plug-and-play disparity confidence estimation algorithm and two depth prior-guided loss functions, and achieves state-of-the-art stereo matching accuracy on the KITTI Stereo benchmarks.
- 4
- 4
- A Robust Long-Term Pedestrian Tracking-by-Detection Algorithm Based on Three-Way Decision4
A new model is proposed that combines the target tracking and target detection and introduces the idea of granular computing to realize high-precision long-term robust pedestrian tracking and significantly improve the tracking accuracy.
- A self-adaptive cascade ConvNets model based on label relation mining4
A Correcting Reliability Level (CRL) supervised three-way decision (3WD) cascade model to implement image classification tasks and introduces two Class Grouping methods to mining the relation between labels, which help to train several expert ConvNets for different types of images.
- Chinese Emotion Recognition Based on Three-Way Decisions4
The results of experiments show that the method of Chinese emotion recognition, based on three-way decisions, has achieved excellent results in the emotion recognition of Chinese sentences.
- 3
- 3
- Text-promptable Object Counting via Quantity Awareness Enhancement3
QUANet is proposed, which introduces novel quantity-oriented text prompts with a vision-text quantity alignment loss to enhance the model's quantity awareness and demonstrates the model's strong generalizability for zero-shot class-agnostic counting.
- 3
- HPT++: Hierarchically Prompting Vision-Language Models with Multi-Granularity Knowledge Generation and Improved Structure Modeling3
This paper proposes a novel approach called Hierarchical Prompt Tuning (HPT), enabling simultaneous modeling of both structured and conventional linguistic knowledge, and introduces a relationship-guided attention module to capture pair-wise associations among entities and attributes for low-level prompt learning.
- DiffPhysBA: Diffusion-based Physical Backdoor Attack against Person Re-Identification in Real-World3
A novel diffusion-based physical backdoor attack (DiffPhysBA) method that adopts a training-free similarity-guided sampling process to enhance the resemblance between generated and physical triggers and can generate realistic attributes as semantic-level triggers in the digital domain.
- 3
- Candidate region acquisition optimization algorithm based on multi-granularity data enhancement3
A multi-granularity selective search algorithm (MGSS) for candidate region acquisition is proposed by extracting the main features such as outline, texture and color of images with multiple grain sizes and improving the subgraph similarity calculation method.
- A Self-adaptive Cascade ConvNets Model Based on Three-Way Decision Theory3
A Correcting Reliability Level (CRL) supervised three-way decision (3WD) cascade model to implement image classification of mass commodity data and introduces a Class Grouping algorithm based on feedback to learn the similarity between classes, which help to train several expert ConvNets for different types of commodity images.
- Sparse embedding visual attention system combined with edge information3
A sparse embedding visual attention system combined with edge information, which is described as a hierarchical model in this paper, and results show that this model outperforms other visual attention computational models.
- 2
- 2
- 2
- DROP: Decouple Re-Identification and Human Parsing with Task-specific Features for Occluded Person Re-identification2
The paper introduces the Decouple Re-identificatiOn and human Parsing (DROP) method for occluded person re-identification (ReID), arguing that the inferior performance of the former is due to distinct granularity requirements for ReID and human parsing features.
- Cap2Sum: Learning to Summarize Videos by Generating Captions2
The use of dense video captions as a supervision signal to train video summarization models and a CLIP (a strong vision-language model) Prior mechanism to enhance the learning of important objects that captions may ignore in the videos are introduced.
- 2
- Saliency-Based Person Re-identification by Probability Histogram2
An effective feature representation based on salient regions and a pool of multiple metric learning for person re-identification and Experimental results show that the proposed method outperforms others.
- 2
- Improving Long-Context Summarization with Multi-Granularity Retrieval Optimization1
The Hierarchical Two-Stage Summarization-based Information Retrieval (HTSIR) method is proposed, which preprocesses the corpus prior to retrieval, summarizes continuous texts to obtain integrated information, and constructs a retrieval tree with varying summary granularities.
- One for All: A Review of Large Pre-training Models for Re-Identification1
This paper presents a comprehensive survey of large pre-training models for ReID, referring to as the ''One for All'' approach, which leverages large-scale pre-training models, such as self-supervised or language- image pre-training models, as general feature extractors adaptable to multiple ReID tasks.
- 1
- 1
- One Object, Multiple Lies: A Benchmark for Cross-Task Adversarial Attack on Unified Vision-Language Models1
This paper introduces CrossVLAD, a new benchmark dataset carefully curated from MSCOCO with GPT-4-assisted annotations for systematically evaluating cross-task adversarial attacks on unified VLMs, and proposes a novel success rate metric that measures simultaneous misclassification across all tasks, providing a rigorous evaluation of adversarial transferability.
- Transformer-Based Person Search with High-Frequency Augmentation and Multi-Wave Mixing1
A novel High-frequency Augmentation and Multi-Wave mixing (HAMW) method for person search is proposed, designed to enhance the discriminative feature extraction capabilities of transformers while reducing computational overhead and improving efficiency.
- 1
- 1
- Adaptive Discriminative Regularization for Visual Classification1
This paper proposes a novel regularization that can yield consistent and non-trivial performance improvements in a variety of visual classification tasks and is robust to long-tailed and noisy label data distribution.
- 1
- Path Aggregation and Dual Supervision Network for Scene Text Detection1
To introduce the from coarse to fine detection idea into the one-stage segmentation algorithm, a single-task multi-level supervision method is designed and deformable convolution is used to break through the limits of CNN’s rectangular receptive field, so that it can better adapt to arbitrary shape scene text.
- 1
- Circle-Block Local Binary Patterns for Face Recognition1
Experimental results show the effectiveness of the proposed Circle-Block Local Binary Patterns for face recognition, which was compared with the LBP-based patterns tested on Extended Yale B face image databases.
- Bayes Network Based Collaborating Control Algorithm in Active Multicamera Network with Applications to Object Tracking1
An active multicamera network framework is designed for human detection and tracking by optimizing the cameras collaborating control based on Bayes network to minimize the number of PTZ cameras with control and optimize the cameras' field of view.
- Two-Dimensional Color Uncorrelated Principal Component Analysis for Feature Extraction with Application to Face Recognition1
The proposed 2DCUPCA provided the theoretical foundations analysis and proved the uncor related property between color-based features in theory, which makes it sure that the extracted features directly from three color image matrices will be uncorrelated.
- 1
- 1
- –
- Dual-Phase Visual-Language Pretraining and Adaptation for Long-Tailed Multi-Label Recognition–
DP-VLPA is proposed, a novel Dual-Phase Visual-Language Pretraining and Adaptation framework that employs a Large Language Model to create detailed descriptions that explicitly emphasize tail classes and their contextual relationships, providing a strong and less-biased feature foundation.
- –
- MARLA: An Autonomous Agent for Medical AI Research and Development–
The Medical AI Research Loop Agent (MARLA), an agentic framework that completely abstracts the construction and supervision of medical AI research loops from clinicians, is introduced and its ability to autonomously conduct complex medical AI research while substantially reducing the need for AI expertise is demonstrated.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- TRAIL: Transferable Robust Adversarial Images via Latent diffusion–
TRAIL is proposed, a test-time adaptation framework that enables the model to generate images from a distribution of images with adversarial features and closely resembles the target images, validating that distribution-aligned adversarial feature synthesis is critical for practical black-box attacks.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
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.