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
- Lizhuang MaSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works160 from public data
- 449
- Local Relation Learning for Face Forgery Detection359
This work proposes a Multi-scale Patch Similarity Module (MPSM), which measures the similarity between features of local regions and forms a robust and generalized similarity pattern and an RGB-Frequency Attention Module (RFAM) to fuse information in both RGB and frequency domains for more comprehensive local feature representation, which further improves the reliability of the similarity pattern.
- 316
- 281
- Federated Learning with Label Distribution Skew via Logits Calibration271
This paper investigates the label distribution skew in FL and proposes FedLC, which calibrates the logits before softmax cross-entropy according to the probability of occurrence of each class, and applies a fine-grained calibrated cross-entropy loss to local update by adding a pairwise label margin.
- Dual Contrastive Learning for General Face Forgery Detection260
A novel face forgery detection framework, named Dual Contrastive Learning (DCL), which specially constructs positive and negative paired data and performs designed contrastive learning at different granularities to learn generalized feature representation is proposed.
- Spatiotemporal Inconsistency Learning for DeepFake Video Detection241
A novel temporal modeling paradigm in TIM is presented by exploiting the temporal difference over adjacent frames along with both horizontal and vertical directions and the ISM simultaneously utilizes the spatial information from SIM and temporal information from TIM to establish a more comprehensive spatial-temporal representation.
- 213
- 209
- Adv-Makeup: A New Imperceptible and Transferable Attack on Face Recognition204
A unified adversarial face generation method - Adv-Makeup, which can realize imperceptible and transferable attack under the black-box setting, and implements a fine-grained meta-learning based adversarial attack strategy to learn more vulnerable or sensitive features from various models.
- Exploiting Fine-Grained Face Forgery Clues via Progressive Enhancement Learning194
A fine-grained decomposition of RGB images is performed to completely decouple the real and fake traces in the frequency space and this method outperforms the state-of-the-art face forgery detection methods.
- Face Anti-Spoofing via Disentangled Representation Learning155
A novel perspective of face anti-spoofing is proposed that disentangles the liveness features and content features from images, and theLiveness features is further used for classification, and a Convolutional Neural Network architecture is put forward.
- HDMixer: Hierarchical Dependency with Extendable Patch for Multivariate Time Series Forecasting127
This work designs a Length-Extendable Patcher (LEP) tailored to MTS, which enriches the boundary information of patches and alleviates semantic incoherence in series, and devise a Hierarchical Dependency Explorer (HDE) based on pure MLPs.
- Generalizable Representation Learning for Mixture Domain Face Anti-Spoofing127
This work designs a domain feature based on Instance Normalization and proposes a domain representation learning module (DRLM) to extract discriminative domain features for clustering and designs a generalizable face anti-spoofing with meta-learning to enhance the interpretability through visualization.
- Delving into the Local: Dynamic Inconsistency Learning for DeepFake Video Detection124
This work elaborately design an Intra-Snippet Inconsistency Module (Intra-SIM) and an Inter- Snippet Interaction Module (Inter- SIM) to establish a dynamic inconsistency modelling frame- work and outperforms the state of the art competitors on four popular benchmark dataset.
- Adaptive Normalized Representation Learning for Generalizable Face Anti-Spoofing118
A novel perspective of face anti-spoofing that focuses on the normalization selection in the feature extraction process is proposed, which adaptively selects feature normalization methods according to the inputs, aiming to learn domain-agnostic and discriminative representation.
- 107
- Dual Reweighting Domain Generalization for Face Presentation Attack Detection106
A novel Dual Reweighting Domain Generalization (DRDG) framework which iteratively reweights the relative importance between samples to further improve the generalization.
- Instance-Aware Domain Generalization for Face Anti-Spoofing105
Instance-Aware Domain Generalization framework is proposed to learn the generalizable feature by weakening the features' sensitivity to instance-specific styles by adaptingively eliminate the style-sensitive feature correlation, boosting the generalization.
- Exploring Frequency Adversarial Attacks for Face Forgery Detection101
This paper proposes a frequency adversarial attack method against face forgery detectors that fools not only the spatial-based detectors but also the state-of- the-art frequency-based detector effectively and enhances the transferability across face forgeries detectors as black-box attacks.
- Generalizing Deepfake Video Detection with Plug-and-Play: Video-Level Blending and Spatiotemporal Adapter Tuning89
This paper investigates whether and how video-level blending can be effective in video and designs a lightweight Spatiotemporal Adapter (StA) to equip a pretrained image model with the ability to capture both spatial and temporal features jointly and efficiently.
- Disentangled High Quality Salient Object Detection89
A novel deep learning framework for high- resolution SOD task is proposed, which disentangles the task into a low-resolution saliency classification network (LRSCN) and a high-resolution refinement network (HRRN), which achieves superior performance compared to the state-of-the-art methods.
- LaRE 2 : Latent Reconstruction Error Based Method for Diffusion-Generated Image Detection83
This work comes up with the Latent Reconstruction Error (LaRE), the first reconstruction-error based feature in the latent space for generated image detection and proposes an Error-Guided feature REfinement module (EGRE), which can refine the image feature guided by LaRE to enhance the discriminativeness of the feature.
- 83
- Towards Practical Certifiable Patch Defense with Vision Transformer78
A progressive smoothed image modeling task to train Vision Transformer, which can capture the more discriminable local context of an image while preserving the global semantic information and obtains state-of-the-art clean and certified accuracy with inferring efficiently on CIFAR-10 and ImageNet.
- Shape Matters: Deformable Patch Attack77
A novel De-formable Patch Representation (DPR) is proposed that can harness the geometric structure of triangles to support the differentiable mapping between contour modeling and masks and achieves state-of-the-art attack performance by deforming shapes on GTSRB and ILSVRC2012.
- Adaptive Mixture of Experts Learning for Generalizable Face Anti-Spoofing76
An Adaptive Mixture of Experts Learning (AMEL) framework is proposed, which exploits the domain-specific information to adaptively establish the link among the seen source domains and unseen target domains to further improve the generalization.
- Generative Domain Adaptation for Face Anti-Spoofing75
A novel perspective of UDA FAS is proposed that directly fits the target data to the models, and further feeds the stylized data into the well-trained source model for classification, and combines two carefully designed consistency constraints.
- Delving into Data: Effectively Substitute Training for Black-box Attack73
A novel perspective substitute training that focuses on designing the distribution of data used in the knowledge stealing process and the combination of these two modules can further boost the consistency of the substitute model and target model, which greatly improves the effectiveness of adversarial attack.
- Model Tailor: Mitigating Catastrophic Forgetting in Multi-modal Large Language Models69
This paper presents a comprehensive analysis of catastrophic forgetting in MLLMs and introduces a post-training adjustment method called Model Tailor, which primarily preserves the pre-trained parameters while replacing a small number of fine-tuned parameters.
- Sibling-Attack: Rethinking Transferable Adversarial Attacks against Face Recognition66
Sibling-Attack is presented, a new FR attack technique for the first time that explores a novel multi-task perspective and outperforms state-of-the-art FR attack techniques by a non-trivial margin.
- 65
- 65
- Test-Time Domain Generalization for Face Anti-Spoofing63
A novel Test-Time Domain Generalization (TTDG) framework for FAS is introduced, which leverages the testing data to boost the model's generalizability and can be seamlessly integrated into not only the CNN but also ViT backbones.
- Privacy-Preserving Face Recognition Using Trainable Feature Subtraction59
Inspired by image compression, this work proposes creating a visually uninformative face image through feature subtraction between an original face and its model-produced regeneration, resulting in randomized recognizable images devoid of attacker-leverageable texture details.
- 57
- Query-Efficient Decision-Based Black-Box Patch Attack57
This work proposes a differential evolutionary algorithm named DevoPatch, which outperforms the state-of-the-art black-box patch attacks in terms of patch area and attack success rate within a given query budget on image classification and face verification, and can evaluate the robustness of models to black- box patch attacks.
- Orthogonal Subspace Decomposition for Generalizable AI-Generated Image Detection51
A novel approach to incorporate the rich semantic knowledge embedded within large-scale vision foundation models to expand the previous discriminative space (based on forgery patterns only), such that the discrimination is decided by both forgery and semantic cues, thereby reducing the overfitting to specific forgery patterns.
- ContrastMask: Contrastive Learning to Segment Every Thing51
This paper proposes a new method, named ContrastMask, which learns a mask segmentation model on both base and novel categories under a unified pixel-level contrastive learning framework, and demonstrates the superiority of this method, which outperforms previous state-of-the-arts methods.
- Feature Generation and Hypothesis Verification for Reliable Face Anti-spoofing50
A feature generation networks which generate hypotheses of real faces and known attacks are introduced for the first time in the FAS task and Bayesian uncertainty estimation is given, which provides theoretical support for reliable defense in unknown domains.
- DuetFace: Collaborative Privacy-Preserving Face Recognition via Channel Splitting in the Frequency Domain49
Results show that the proposed DuetFace achieves a comparable recognition accuracy and computation cost to the unprotected ArcFace and outperforms the state-of-the-art privacy-preserving methods.
- Structure Destruction and Content Combination for Face Anti-Spoofing48
The former mechanism destroys images into patches to construct a non-structural input, while the latter mechanism recombines patches from different subdomains or classes into a mixup construct based on this splitting-and-splicing operation.
- Contrastive Pseudo Learning for Open-World DeepFake Attribution47
A novel framework named Contrastive Pseudo Learning (CPL) is proposed for the OW-DFA task through introducing a Global-Local Voting module to guide the feature alignment of forged faces with different manipulated regions, and designing a Confidence-based Soft Pseudo-label strategy to mitigate the pseudo-noise caused by similar methods in unlabeled set.
- 47
- 45
- 45
- 45
- Aigi-Holmes: Towards Explainable and Generalizable AI-Generated Image Detection via Multimodal Large Language Models44
The Holmes Pipeline adapts multimodal large language models (MLLMs) for AIGI detection while generating human-verifiable and human-aligned explanations, ultimately yielding the model AIGI-Holmes, which is validated on three benchmarks.
- DistilPose: Tokenized Pose Regression with Heatmap Distillation44
A novel human pose estimation framework termed DistilPose, which bridges the gaps between heatmap-based and regression-based methods and maximizes the transfer of knowledge from the teacher model to the student model through Token-distilling Encoder (TDE) and Simulated Heatmaps.
- Evaluation-oriented Knowledge Distillation for Deep Face Recognition43
A novel Evaluation-oriented KD method (EKD) for deep face recognition is proposed to directly reduce the performance gap between the teacher and student models during training and demonstrates the superiority of the EKD over state-of-the-art competitors.
- SoccerNet 2022 Challenges Results42
The SoccerNet 2022 challenges were the second annual video understanding challenges organized by the SoccerNet team, and tasks (3-6) were novel, including their underlying data and annotations.
- Seeing in Flowing: Adapting CLIP for Action Recognition with Motion Prompts Learning40
This work proposes a two-stream motion modeling block to capture motion and spatial information at the same time and achieves competitive performance on "closed-set" training with extremely few trainable parameters and additional computational costs.
- 38
- Second Edition FRCSyn Challenge at CVPR 2024: Face Recognition Challenge in the Era of Synthetic Data37
An overview of the 2nd edition of the Face Recognition Challenge in the Era of Synthetic Data (FRCSyn) organized at CVPR 2024 is presented, which proposes new sub-tasks that allow participants to explore novel face generative methods.
- Privacy-Preserving Face Recognition Using Random Frequency Components37
An in-depth study of the privacy protection of face images’ visual information and against recovery, and proposes a solution to the dilemma, by advocating for the training and inference of recognition models on randomly selected frequency components.
- Continual Face Forgery Detection via Historical Distribution Preserving35
A Historical Distribution Preserving (HDP) framework is proposed that reserves and preserves the distributions of historical faces and uses universal adversarial perturbation to simulate historical forgery distribution and knowledge distillation to maintain the distribution variation of real faces across different models.
- Face Manipulation Detection via Auxiliary Supervision35
A multi-scale framework is proposed to estimate the manipulation probability with texture map and blending boundary as auxiliary supervisions to guide the network to learn discriminative and generalizable cues.
- ID$^3$: Identity-Preserving-yet-Diversified Diffusion Models for Synthetic Face Recognition34
A diffusion-fueled SFR model that employs an ID-preserving loss to generate diverse yet identity-consistent facial appearances is introduced, and it is shown that minimizing this loss is equivalent to maximizing the lower bound of an adjusted conditional log-likelihood over ID-preserving data.
- SoccerNet 2023 challenges results34
This report indicates performance trends across tasks: Action spotting is nearing saturation, while ball action spotting improved significantly with advanced end-to-end models and dense video captioning also saw substantial enhancements aligned with Large Language Models advancements.
- Exploiting the Textual Potential from Vision-Language Pre-training for Text-based Person Search32
This paper builds on the proposed VLP-TPS baseline model, which is the first TPS model with both pre-trained modalities, and proposes the Multi-Integrity Description Constraints (MIDC) to enhance the robustness of the textual modality by incorporating different components of fine-grained corpus during training.
- Region-Aware Temporal Inconsistency Learning for DeepFake Video Detection29
A novel Region-Aware Temporal Filter (RATF) module which automatically generates corresponding temporal filters for different spatial regions and guides the corresponding regions to adaptively learn temporal inconsistency, which greatly enhances the overall representational ability.
- Towards Unified Defense for Face Forgery and Spoofing Attacks via Dual Space Reconstruction Learning28
A dual space reconstruction learning framework that models the commonalities of genuine faces in both spatial and frequency domains and introduces a dynamic filtering module that filters out the redundant information retained by the reconstruction and enhances the critical divergence between the real and the attack to achieve better classification features.
- 27
- 26
- BézierPalm: A Free Lunch for Palmprint Recognition25
An intuitive geometric model which represents palmar creases with parameterized B´ezier curves is introduced, which can synthesize massive training samples of diverse identities, which enables us to pretrain large-scale palmprint recognition models.
- Second FRCSyn-onGoing: Winning solutions and post-challenge analysis to improve face recognition with synthetic data24
The 2nd FRCSyn-onGoing challenge is introduced, based on the 2nd Face Recognition Challenge in the Era of Synthetic Data, and focuses on exploring the use of synthetic data both individually and in combination with real data to solve current challenges in face recognition such as demographic bias, domain adaptation, and performance constraints in demanding situations.
- 24
- MS-DETR: Towards Effective Video Moment Retrieval and Highlight Detection by Joint Motion-Semantic Learning23
This paper proposes the Motion-Semantics DETR (MS-DETR), a framework that captures rich motion-semantics features through unified learning for MR/HD tasks and outperforms existing state-of-the-art models by a margin.
- 23
- Toward Stable Co-Saliency Detection and Object Co-Segmentation22
The proposed MSRU not only helps CoSOD (CoSEG) model captures robust inter-image relations, but also reduces order-sensitivity, resulting in a more stable inference and training process, and design a cross-order contrastive loss (COCL) that can further address order-sensitive problem by pulling close the feature embedding generated from different input orders.
- Rethinking Open-World DeepFake Attribution with Multi-perspective Sensory Learning21
A novel task named Open-World DeepFake Attribution, and the corresponding benchmark OW-DFA++, which aims to evaluate attribution performance against various types of fake faces in open-world scenarios are introduced and a Multi-Perspective Sensory Learning (MPSL) framework is proposed that aims to address the challenge of OW-DFA++.
- Attack Can Benefit: An Adversarial Approach to Recognizing Facial Expressions under Noisy Annotations21
A novel and flexible method to spot noisy labels by leveraging adversarial attack, termed as Geometry Aware Adversarial Vulnerability Estimation (GAAVE), which has no reliance on additional information and is thus easy to generalize to the large-scale real-world FER datasets.
- RPG-Palm: Realistic Pseudo-data Generation for Palmprint Recognition21
This paper proposes a novel realistic pseudo-palmprint generation (RPG) model to synthesize palmprints with massive identities, and introduces a conditional modulation generator to improve the intra-class diversity and an identity-aware loss to ensure identity consistency against unpaired training.
- 21
- PCE-Palm: Palm Crease Energy Based Two-Stage Realistic Pseudo-Palmprint Generation20
A new palm crease energy (PCE) domain is introduced as a bridge from Bézier curves to real palmprints and a two-stage generation model is proposed, achieving a 29% improvement over RPG-Palm and outperforms ArcFace with 100% training set by more than 6% in terms of TAR@FAR=1e-6.
- 20
- Highly Efficient Natural Image Matting20
An extremely light-weighted model is constructed, which achieves comparable performance with ~1\% parameters (344k) of large models on popular natural image matting benchmarks.
- Detecting Adversarial Patch Attacks through Global-local Consistency20
This paper proposes a simple but very effective approach to detect adversarial patches based on an interesting observation called global-local consistency and proposes to use Random-Local-Ensemble strategy to further enhance it in the detection.
- 19
- 18
- TF-FAS: Twofold-Element Fine-Grained Semantic Guidance for Generalizable Face Anti-spoofing18
This work proposes a novel framework called TF-FAS, which aims to thoroughly explore and harness twofold-element fine-grained semantic guidance to enhance generalization, and presents a Fine-Grained Categorical Element Module (FCEM) to explore fine-grained categorical element guidance.
- Anchor-based Robust Finetuning of Vision-Language Models18
This work proposes to compensate for the finetune process using auxiliary supervision with rich semantic information, which acts as anchors to preserve the OOD generalization in CLIP, thereby preserving the OOD generalization capabilities.
- Edge-preserving image decomposition via joint weighted least squares18
Experimental results indicate that the proposed approach can appropriately smooth non-edge regions even when textures and structures are similar in scale, and the effectiveness of the approach is demonstrated in the contexts of detail manipulation, HDR tone mapping, and image abstraction.
- Adma-GAN: Attribute-Driven Memory Augmented GANs for Text-to-Image Generation.17
This work constructs an attribute memory to jointly control the text-to-image generation with sentence input and designs an attribute-sentence-joint conditional generator learning scheme to align the feature embeddings among multiple representations, which promotes the cross-modal network training.
- 17
- Data Synthesis with Diverse Styles for Face Recognition via 3DMM-Guided Diffusion16
To create virtual faces, the generator is conditioned on novel identities of unlabeled synthetic faces, and novel styles that are statistically sampled from a real-world prior distribution, which accounts for both intra-subject variation and subject distinctiveness.
- Instruct Where the Model Fails: Generative Data Augmentation via Guided Self-contrastive Fine-tuning16
The experiments on few-shot class incremental learning show that the instruction-guided finetuning strategy consistently assists the downstream model with higher classification accuracy compared to generative data augmentation baselines such as Stable Diffusion and GPT-4o, and state-of-the-art non-generative strategies.
- SDPose: Tokenized Pose Estimation via Circulation-Guide Self-Distillation15
SD-Pose, a new self-distillation method for improving the performance of small transformer-based models, is introduced based on a transformer module named Multi-Cycled Transformer based on multiple-cycled forwards to more fully exploit the potential of small model parameters.
- Antidote: A Unified Framework for Mitigating LVLM Hallucinations in Counterfactual Presupposition and Object Perception14
This paper introduces "Antidote", a unified, synthetic data-driven post-training framework for mitigating both types of hallucination above, which leverages synthetic data to incorporate factual priors into questions to achieve self-correction, and decouple the mitigation process into a preference optimization problem.
- 14
- 13
- 12
- 12
- 12
- 11
- 11
- 11
- Similarity metric learning for face verification using sigmoid decision function11
The proposed face verification method is based on the idea that it would like the similarity between positive pairs larger than negative pairs, and obtain a similarity estimation of two images, and belongs to nonlinear metric which is more robust to handle heterogeneous data than linear metric.
- 10
- Decoupled Data Augmentation for Improving Image Classification10
Decoupled Data Augmentation (De-DA), which resolves the fidelity-diversity dilemma by separating images into CDPs and CIPs and handling them adaptively, and replaces the image's CIP with inter-class variants, creating diverse CDP-CIP combinations.
- 10
- A Quality-Centric Framework for Generic Deepfake Detection9
A novel quality-centric framework for generic deepfake detection is proposed, which is composed of a Quality Evaluator, a low-quality data enhancement module, and a learning pacing strategy that explicitly incorporates forgery quality into the training process.
- 8
- 8
- 8
- 8
- Intrinsic image estimation using near- $$L_0$$ L 0 sparse optimization8
A novel approach for automatic intrinsic images decomposition that uses a new reflectance sparsity prior that significantly improves the quality of Retinex intrinsic images estimation and functions effectively by combining a class of global sparsity priors on reflectance.
- ToVE: Efficient Vision-Language Learning via Knowledge Transfer from Vision Experts7
This paper proposes a new framework that Transfers the knowledge from a hub of Vision Experts (ToVE) for efficient VL learning, leveraging pre-trained vision expert models to promote visual perception capability.
- 7
- 7
- From Enhancement to Understanding: Build a Generalized Bridge for Low-Light Vision via Semantically Consistent Unsupervised Fine-Tuning6
This work introduces an illumination-aware image prompt to explicitly guide image generation and proposes a cycle-attention adapter to maximize its semantic potential and introduces Semantically Consistent Unsupervised Fine-tuning (SCUF), a paradigm that improves both generalization and scalability.
- Diff-Palm: Realistic Palmprint Generation with Polynomial Creases and Intra-Class Variation Controllable Diffusion Models6
Experimental results show that, for the first time, recognition models trained solely on synthetic datasets, without any fine-tuning, outperform those trained on real datasets and achieves superior recognition performance as the number of generated identities increases.
- Towards Generalizable Data Protection With Transferable Unlearnable Examples6
This paper presents a novel, generalizable data protection method by generating transferable unlearnable examples, and is the first solution that examines data privacy from the perspective of data distribution.
- 6
- Eigen-Aging Reference Coding for Cross-Age Face Verification and Retrieval6
A new reference based approach to address cross-age face recognition problem, called Eigen-Aging Reference Coding (EARC), which achieves state-of-the-art performance and speed on CACD dataset, the largest public face dataset containing significant aging information.
- Fuse Before Transfer: Knowledge Fusion for Heterogeneous Distillation5
This fusion combines the advantages of both cross-architecture inductive biases and module functions by merging different combinations of convolution, attention, and MLP modules derived directly from student and teacher module functions.
- 5
- DITL 2 : Dual-Stage Invariance Transfer Learning for Generalizable Document Image Tampering Localization5
DITL2 is proposed, a Dual-stage Invariance Transfer Learning framework for Document Image Tampering Localization that consists of Cross-Domain Invariance Pre-training (CDIP) and Frequency Decoupling Parameter Adaptation (FDPA).
- 5
- Intrinsic Image Decomposition Using Multi‐Scale Measurements and Sparsity5
A new algorithm is proposed that generates and combines multi‐scale properties of chromaticity differences and intensity contrast and can be solved efficiently within a few iterations of the problem through energy minimization.
- 4
- UIFace: Unleashing Inherent Model Capabilities to Enhance Intra-Class Diversity in Synthetic Face Recognition4
A framework to Unleash Inherent capability of the model to enhance intra-class diversity for synthetic face recognition, shortened as UIFace is proposed and significantly surpasses previous approaches with even less training data and half the size of synthetic dataset.
- Domain-Hallucinated Updating for Multi-Domain Face Anti-spoofing4
This work proposes a novel Domain-Hallucinated Updating (DHU) framework to facilitate the hallucination of data for updating FAS model and demonstrates that the proposed method outperforms state-of-the-art competitors in terms of effectiveness.
- Bilateral Adaptive Cross-Modal Fusion Prompt Learning for CLIP4
This work proposes a new paradigm for the CLIP prompt learning community, named Bilateral Adaptive Cross-Modal Fusion Prompt Learning (Bloom), which includes two enhancements, including an adaptive manner that automatically searches the optimal combination of cross-modal information at each layer.
- 4
- 4
- 3
- Generalizable Person Search on Open-world User-Generated Video Content3
This work focuses on learning domain-invariant representations for both detection and ReID by introducing a multi-task prototype-based domain-specific batch normalization, and a channel-wise ID-relevant feature decorrelation strategy.
- 2
- Exploring the Collaborative Advantage of Low-level Information on Generalizable AI-Generated Image Detection2
This work introduces Lora Experts, enabling the backbone network, which is trained with high-level semantic RGB images, to accept and learn knowledge from different low-level information, and proposes Dynamic Feature Selection, which dynamically selects the most suitable features for detecting the current image to maximize generalization detection capability.
- 2
- 2
- GloTok: Global Perspective Tokenizer for Image Reconstruction and Generation2
A Global Perspective Tokenizer (GloTok), which utilizes global relational information to model a more uniform semantic distribution of tokenized features, which facilitates the training of autoregressive (AR) models for generating high-quality images without requiring direct access to pre-trained models during the training process.
- AlignCLIP: Align Multi Domains of Texts Input for CLIP models with Object-IoU Loss2
This paper proposes the use of caption object parsing to identify the objects set contained within captions, and demonstrates that this method significantly surpasses the CLIP baseline and exceeds existing methods, achieving a new state-of-the-art (SOTA).
- Dual Relation Mining Network for Zero-Shot Learning2
This work introduces a Dual Attention Block (DAB) for visual-semantic relationship mining, which enriches visual information by multi-level feature fusion and conducts spatial attention for visual to semantic embedding and an attribute-guided channel attention is utilized to decouple entangled semantic features.
- Re-thinking Data Availablity Attacks Against Deep Neural Networks2
This paper re-examines the concept of unlearnable examples and introduces a novel optimization paradigm that yields improved protection results with reduced computational time requirements and establishes a robust foundation for future research in this area.
- Combining Past, Present and Future: A Self-Supervised Approach for Class Incremental Learning2
A self-supervised CIL framework CPPF, meaning Combining Past, Present and Future, is proposed, which boosts the performance of self- supervised class incremental learning.
- Integrated tone and structure refinement for high‐fidelity colour transfer2
The authors develop the ideas of image decomposition and gradient guidance to perform tone reconstruction while protecting original structure while demonstrating the performance of the proposed method through a number of experiments in visual comparison and objective evaluation.
- A Novel Customized Recompression Framework for Massive Internet Images2
This paper proposes a novel framework to efficiently customize image recompression according to a variety of applications, and this new framework has been successfully applied to many commercial applications, such as web portals, e-commerce, online game and so on.
- A Customized Framework to Recompress Massive Internet Images2
A novel framework to efficiently customize image recompression according to a variety of applications is proposed and has been successfully applied to many commercial applications, such as web portals, e-commerce, online game, and so on.
- LSAP-PV: High-Fidelity Palm Vein Image Synthesis via Layered Spectral Absorption Projection-Guided Diffusion Model1
This work proposes a novel biophysics-driven synthesis method that outperforms existing methods, but also enables a recognition model trained on real-world data to achieve superior performance compared to a model trained on real-world data under an open-set protocol.
- 1
- 1
- 1
- Geometric Synthesis: A Free lunch for Large-scale Palmprint Recognition Model Pretraining1
An intuitive geometric model is introduced which represents palmar creases with parameterized B\'ezier curves which enables us to pretrain large-scale palmprint recognition models and demonstrates that such synthetically pretrained models have a very strong generalization ability.
- 1
- 1
- –
- –
- –
- –
- SonarGuard2: Ultrasonic Face Liveness Detection Based on Adaptive Doppler Effect Feature Extraction–
A novel framework, named SonarGuard2, adaptively selecting the ultrasonic signals and analyse the Doppler effect, introduces acoustic echo cancellation to filter to get the Doppler effect features, and utilizes complex convolutional neural networks to enhance the abilitity to model the Doppler effect features.
- –
- –
- Artificial Intelligence Security Competition (AISC)–
The competition rules of these three tracks and the solutions of top-ranking teams in each track are introduced, including Deepfake Security Competition, Autonomous driving security Competition, and Face Recognition Security Competition.
- Anomaly Detection by Latent Regularized Dual Adversarial Networks–
A novel adversarial dual autoencoder network is proposed, in which the underlying structure of training data is not only captured in latent feature space, but also can be further restricted in the space of latent representation in a discriminant manner, leading to a more accurate detector.
- Object-Level Salience Detection by Progressively Enhanced Network–
This framework is composed of two main components: progressively enhanced network (PEN) for amplifying the specified layers of the network and merging the global context simultaneously; (2) object-level boundary extraction module (OBEM) for extracting the complete boundary of the salient object.
- –
- Illumination invariant mesh coloring–
This paper presents a non-photorealistic per-vertex illumination invariant mesh coloring method to couple intrinsic color information into mesh, which can reflect material properties more accurately.
- A Visual Lossless Image-Recompression Framework–
A novel image recompression framework and image quality assessment method to efficiently recompress Internet images is proposed and the file size of images can be reduced by about 45%- 60% without affecting their visual quality.
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 2026-10-14. 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.