Zhenan Sun

Boqiang Xu, Lingxiao He, Jian Liang et al.

One major issue that challenges person re-identification (Re-ID) is the ubiquitous occlusion over the captured persons. There are two main challenges for the occluded person Re-ID problem, i.e., the interference of noise during feature matching and the loss of pedestrian information brought by the occlusions. In this paper, we propose a new approach called Feature Recovery Transformer (FRT) to address the two challenges simultaneously, which mainly consists of visibility graph matching and feature recovery transformer. To reduce the interference of the noise during feature matching, we mainly focus on visible regions that appear in both images and develop a visibility graph to calculate the similarity. In terms of the second challenge, based on the developed graph similarity, for each query image, we propose a recovery transformer that exploits the feature sets of its $k$-nearest neighbors in the gallery to recover the complete features. Extensive experiments across different person Re-ID datasets, including occluded, partial and holistic datasets, demonstrate the effectiveness of FRT. Specifically, FRT significantly outperforms state-of-the-art results by at least 6.2\% Rank-1 accuracy and 7.2\% mAP scores on the challenging Occluded-Duke dataset. The code is available at https://github.com/xbq1994/Feature-Recovery-Transformer.

Qi Li, Weining Wang, Chengzhong Xu et al.

Although face swapping has attracted much attention in recent years, it remains a challenging problem. Existing methods leverage a large number of data samples to explore the intrinsic properties of face swapping without considering the semantic information of face images. Moreover, the representation of the identity information tends to be fixed, leading to suboptimal face swapping. In this paper, we present a simple yet efficient method named FaceSwapper, for one-shot face swapping based on Generative Adversarial Networks. Our method consists of a disentangled representation module and a semantic-guided fusion module. The disentangled representation module comprises an attribute encoder and an identity encoder, which aims to achieve the disentanglement of the identity and attribute information. The identity encoder is more flexible, and the attribute encoder contains more attribute details than its competitors. Benefiting from the disentangled representation, FaceSwapper can swap face images progressively. In addition, semantic information is introduced into the semantic-guided fusion module to control the swapped region and model the pose and expression more accurately. Experimental results show that our method achieves state-of-the-art results on benchmark datasets with fewer training samples. Our code is publicly available at https://github.com/liqi-casia/FaceSwapper.

Min Ren, Yuhao Zhu, Yunlong Wang et al.

Deep learning-based face recognition models are vulnerable to adversarial attacks. To curb these attacks, most defense methods aim to improve the robustness of recognition models against adversarial perturbations. However, the generalization capacities of these methods are quite limited. In practice, they are still vulnerable to unseen adversarial attacks. Deep learning models are fairly robust to general perturbations, such as Gaussian noises. A straightforward approach is to inactivate the adversarial perturbations so that they can be easily handled as general perturbations. In this paper, a plug-and-play adversarial defense method, named perturbation inactivation (PIN), is proposed to inactivate adversarial perturbations for adversarial defense. We discover that the perturbations in different subspaces have different influences on the recognition model. There should be a subspace, called the immune space, in which the perturbations have fewer adverse impacts on the recognition model than in other subspaces. Hence, our method estimates the immune space and inactivates the adversarial perturbations by restricting them to this subspace. The proposed method can be generalized to unseen adversarial perturbations since it does not rely on a specific kind of adversarial attack method. This approach not only outperforms several state-of-the-art adversarial defense methods but also demonstrates a superior generalization capacity through exhaustive experiments. Moreover, the proposed method can be successfully applied to four commercial APIs without additional training, indicating that it can be easily generalized to existing face recognition systems. The source code is available at https://github.com/RenMin1991/Perturbation-Inactivate

Jie Gui, Tuo Chen, Jing Zhang et al.

Deep supervised learning algorithms typically require a large volume of labeled data to achieve satisfactory performance. However, the process of collecting and labeling such data can be expensive and time-consuming. Self-supervised learning (SSL), a subset of unsupervised learning, aims to learn discriminative features from unlabeled data without relying on human-annotated labels. SSL has garnered significant attention recently, leading to the development of numerous related algorithms. However, there is a dearth of comprehensive studies that elucidate the connections and evolution of different SSL variants. This paper presents a review of diverse SSL methods, encompassing algorithmic aspects, application domains, three key trends, and open research questions. Firstly, we provide a detailed introduction to the motivations behind most SSL algorithms and compare their commonalities and differences. Secondly, we explore representative applications of SSL in domains such as image processing, computer vision, and natural language processing. Lastly, we discuss the three primary trends observed in SSL research and highlight the open questions that remain. A curated collection of valuable resources can be accessed at https://github.com/guijiejie/SSL.

Jawad Muhammad, Yunlong Wang, Junxing Hu et al.

Iris biometrics is a phenotypic biometric trait that has proven to be agnostic to human natural physiological changes. Research on iris biometrics has progressed tremendously, partly due to publicly available iris databases. Various databases have been available to researchers that address pressing iris biometric challenges such as constraint, mobile, multispectral, synthetics, long-distance, contact lenses, liveness detection, etc. However, these databases mostly contain subjects of Caucasian and Asian docents with very few Africans. Despite many investigative studies on racial bias in face biometrics, very few studies on iris biometrics have been published, mainly due to the lack of racially diverse large-scale databases containing sufficient iris samples of Africans in the public domain. Furthermore, most of these databases contain a relatively small number of subjects and labelled images. This paper proposes a large-scale African database named CASIA-Iris-Africa that can be used as a complementary database for the iris recognition community to mediate the effect of racial biases on Africans. The database contains 28,717 images of 1023 African subjects (2046 iris classes) with age, gender, and ethnicity attributes that can be useful in demographically sensitive studies of Africans. Sets of specific application protocols are incorporated with the database to ensure the database's variability and scalability. Performance results of some open-source SOTA algorithms on the database are presented, which will serve as baseline performances. The relatively poor performances of the baseline algorithms on the proposed database despite better performance on other databases prove that racial biases exist in these iris recognition algorithms. The database will be made available on our website: http://www.idealtest.org.

Hongwen Zhang, Yating Tian, Yuxiang Zhang et al.

We present PyMAF-X, a regression-based approach to recovering parametric full-body models from monocular images. This task is very challenging since minor parametric deviation may lead to noticeable misalignment between the estimated mesh and the input image. Moreover, when integrating part-specific estimations into the full-body model, existing solutions tend to either degrade the alignment or produce unnatural wrist poses. To address these issues, we propose a Pyramidal Mesh Alignment Feedback (PyMAF) loop in our regression network for well-aligned human mesh recovery and extend it as PyMAF-X for the recovery of expressive full-body models. The core idea of PyMAF is to leverage a feature pyramid and rectify the predicted parameters explicitly based on the mesh-image alignment status. Specifically, given the currently predicted parameters, mesh-aligned evidence will be extracted from finer-resolution features accordingly and fed back for parameter rectification. To enhance the alignment perception, an auxiliary dense supervision is employed to provide mesh-image correspondence guidance while spatial alignment attention is introduced to enable the awareness of the global contexts for our network. When extending PyMAF for full-body mesh recovery, an adaptive integration strategy is proposed in PyMAF-X to produce natural wrist poses while maintaining the well-aligned performance of the part-specific estimations. The efficacy of our approach is validated on several benchmark datasets for body, hand, face, and full-body mesh recovery, where PyMAF and PyMAF-X effectively improve the mesh-image alignment and achieve new state-of-the-art results. The project page with code and video results can be found at https://zhanghongwen.cn/pymaf-x.

Bo Peng, Wei Xiang, Yue Jiang et al.

This paper presents the summary report on our DFGC 2022 competition. The DeepFake is rapidly evolving, and realistic face-swaps are becoming more deceptive and difficult to detect. On the contrary, methods for detecting DeepFakes are also improving. There is a two-party game between DeepFake creators and defenders. This competition provides a common platform for benchmarking the game between the current state-of-the-arts in DeepFake creation and detection methods. The main research question to be answered by this competition is the current state of the two adversaries when competed with each other. This is the second edition after the last year's DFGC 2021, with a new, more diverse video dataset, a more realistic game setting, and more reasonable evaluation metrics. With this competition, we aim to stimulate research ideas for building better defenses against the DeepFake threats. We also release our DFGC 2022 dataset contributed by both our participants and ourselves to enrich the DeepFake data resources for the research community (https://github.com/NiCE-X/DFGC-2022).

Zhengquan Luo, Yunlong Wang, Zilei Wang et al.

Attributes skew hinders the current federated learning (FL) frameworks from consistent optimization directions among the clients, which inevitably leads to performance reduction and unstable convergence. The core problems lie in that: 1) Domain-specific attributes, which are non-causal and only locally valid, are indeliberately mixed into global aggregation. 2) The one-stage optimizations of entangled attributes cannot simultaneously satisfy two conflicting objectives, i.e., generalization and personalization. To cope with these, we proposed disentangled federated learning (DFL) to disentangle the domain-specific and cross-invariant attributes into two complementary branches, which are trained by the proposed alternating local-global optimization independently. Importantly, convergence analysis proves that the FL system can be stably converged even if incomplete client models participate in the global aggregation, which greatly expands the application scope of FL. Extensive experiments verify that DFL facilitates FL with higher performance, better interpretability, and faster convergence rate, compared with SOTA FL methods on both manually synthesized and realistic attributes skew datasets.

Kaiduo Zhang, Muyi Sun, Jianxin Sun et al.

Text-driven person image generation is an emerging and challenging task in cross-modality image generation. Controllable person image generation promotes a wide range of applications such as digital human interaction and virtual try-on. However, previous methods mostly employ single-modality information as the prior condition (e.g. pose-guided person image generation), or utilize the preset words for text-driven human synthesis. Introducing a sentence composed of free words with an editable semantic pose map to describe person appearance is a more user-friendly way. In this paper, we propose HumanDiffusion, a coarse-to-fine alignment diffusion framework, for text-driven person image generation. Specifically, two collaborative modules are proposed, the Stylized Memory Retrieval (SMR) module for fine-grained feature distillation in data processing and the Multi-scale Cross-modality Alignment (MCA) module for coarse-to-fine feature alignment in diffusion. These two modules guarantee the alignment quality of the text and image, from image-level to feature-level, from low-resolution to high-resolution. As a result, HumanDiffusion realizes open-vocabulary person image generation with desired semantic poses. Extensive experiments conducted on DeepFashion demonstrate the superiority of our method compared with previous approaches. Moreover, better results could be obtained for complicated person images with various details and uncommon poses.

Qi Li, Weining Wang, Shuangjun Du et al.

Face swapping has witnessed significant progress in recent years, largely driven by advances in deep generative models such as GANs and diffusion models.Despite these advances, existing methods remain fragmented across different paradigms, and their evaluation is highly inconsistent due to the lack of standardized datasets and protocols. Moreover, prior surveys primarily focus on broader deepfake generation or detection, leaving face swapping insufficiently studied as a standalone problem. In this paper, we present a comprehensive survey and benchmark for face swapping. We provide a structured review of existing methods, organizing them into five major paradigms and systematically analyzing their design principles, strengths, and limitations. To enable fair and controlled evaluation, we introduce CASIA FaceSwapping, a high-quality benchmark with balanced demographic distributions and explicit attribute variations, and establish standardized protocols to assess the robustness of different face swapping methods. Extensive experiments on representative approaches yield new insights into the performance characteristics and limitations of current techniques. Overall, our work provides a unified perspective and a principled evaluation framework to facilitate the development of more robust and controllable face swapping methods. More results can be found at https://github.com/CASIA-NLPRAI/face-swapping-survey.

Hongda Liu, Yunlong Wang, Min Ren et al.

Skeleton-based action recognition has recently made significant progress. However, data imbalance is still a great challenge in real-world scenarios. The performance of current action recognition algorithms declines sharply when training data suffers from heavy class imbalance. The imbalanced data actually degrades the representations learned by these methods and becomes the bottleneck for action recognition. How to learn unbiased representations from imbalanced action data is the key to long-tailed action recognition. In this paper, we propose a novel balanced representation learning method to address the long-tailed problem in action recognition. Firstly, a spatial-temporal action exploration strategy is presented to expand the sample space effectively, generating more valuable samples in a rebalanced manner. Secondly, we design a detached action-aware learning schedule to further mitigate the bias in the representation space. The schedule detaches the representation learning of tail classes from training and proposes an action-aware loss to impose more effective constraints. Additionally, a skip-modal representation is proposed to provide complementary structural information. The proposed method is validated on four skeleton datasets, NTU RGB+D 60, NTU RGB+D 120, NW-UCLA, and Kinetics. It not only achieves consistently large improvement compared to the state-of-the-art (SOTA) methods, but also demonstrates a superior generalization capacity through extensive experiments. Our code is available at https://github.com/firework8/BRL.

Min Ren, Yunlong Wang, Yuhao Zhu et al.

Occlusion is a common problem with biometric recognition in the wild. The generalization ability of CNNs greatly decreases due to the adverse effects of various occlusions. To this end, we propose a novel unified framework integrating the merits of both CNNs and graph models to overcome occlusion problems in biometric recognition, called multiscale dynamic graph representation (MS-DGR). More specifically, a group of deep features reflected on certain subregions is recrafted into a feature graph (FG). Each node inside the FG is deemed to characterize a specific local region of the input sample, and the edges imply the co-occurrence of non-occluded regions. By analyzing the similarities of the node representations and measuring the topological structures stored in the adjacent matrix, the proposed framework leverages dynamic graph matching to judiciously discard the nodes corresponding to the occluded parts. The multiscale strategy is further incorporated to attain more diverse nodes representing regions of various sizes. Furthermore, the proposed framework exhibits a more illustrative and reasonable inference by showing the paired nodes. Extensive experiments demonstrate the superiority of the proposed framework, which boosts the accuracy in both natural and occlusion-simulated cases by a large margin compared with that of baseline methods.

Zhuojie Wu, Xingqun Qi, Zijian Wang et al.

Face inpainting aims to complete the corrupted regions of the face images, which requires coordination between the completed areas and the non-corrupted areas. Recently, memory-oriented methods illustrate great prospects in the generation related tasks by introducing an external memory module to improve image coordination. However, such methods still have limitations in restoring the consistency and continuity for specificfacial semantic parts. In this paper, we propose the coarse-to-fine Memory-Disentangled Refinement Networks (MDRNets) for coordinated face inpainting, in which two collaborative modules are integrated, Disentangled Memory Module (DMM) and Mask-Region Enhanced Module (MREM). Specifically, the DMM establishes a group of disentangled memory blocks to store the semantic-decoupled face representations, which could provide the most relevant information to refine the semantic-level coordination. The MREM involves a masked correlation mining mechanism to enhance the feature relationships into the corrupted regions, which could also make up for the correlation loss caused by memory disentanglement. Furthermore, to better improve the inter-coordination between the corrupted and non-corrupted regions and enhance the intra-coordination in corrupted regions, we design InCo2 Loss, a pair of similarity based losses to constrain the feature consistency. Eventually, extensive experiments conducted on CelebA-HQ and FFHQ datasets demonstrate the superiority of our MDRNets compared with previous State-Of-The-Art methods.

Hongda Liu, Yunfan Liu, Min Ren et al.

In skeleton-based human activity understanding, existing methods often adopt the contrastive learning paradigm to construct a discriminative feature space. However, many of these approaches fail to exploit the structural inter-class similarities and overlook the impact of anomalous positive samples. In this study, we introduce ACLNet, an Affinity Contrastive Learning Network that explores the intricate clustering relationships among human activity classes to improve feature discrimination. Specifically, we propose an affinity metric to refine similarity measurements, thereby forming activity superclasses that provide more informative contrastive signals. A dynamic temperature schedule is also introduced to adaptively adjust the penalty strength for various superclasses. In addition, we employ a margin-based contrastive strategy to improve the separation of hard positive and negative samples within classes. Extensive experiments on NTU RGB+D 60, NTU RGB+D 120, Kinetics-Skeleton, PKU-MMD, FineGYM, and CASIA-B demonstrate the superiority of our method in skeleton-based action recognition, gait recognition, and person re-identification. The source code is available at https://github.com/firework8/ACLNet.

Haokun Lin, Xinle Jia, Haobo Xu et al.

The MXFP4 microscaling format, which partitions tensors into blocks of 32 elements sharing an E8M0 scaling factor, has emerged as a promising substrate for efficient LLM inference, backed by native hardware support on NVIDIA Blackwell Tensor Cores. However, activation outliers pose a unique challenge under this format: a single outlier inflates the shared block scale, compressing the effective dynamic range of the remaining elements and causing significant quantization error. Existing rotation-based remedies, including randomized Hadamard and learnable rotations, are data-agnostic and therefore unable to specifically target the channels where outliers concentrate. We propose DuQuant++, which adapts the outlier-aware fine-grained rotation of DuQuant to the MXFP4 format by aligning the rotation block size with the microscaling group size (B{=}32). Because each MXFP4 group possesses an independent scaling factor, the cross-block variance issue that necessitates dual rotations and a zigzag permutation in the original DuQuant becomes irrelevant, enabling DuQuant++ to replace the entire pipeline with a single outlier-aware rotation, which halves the online rotation cost while simultaneously smoothing the weight distribution. Extensive experiments on the LLaMA-3 family under MXFP4 W4A4 quantization show that DuQuant++ consistently achieves state-of-the-art performance. Our code is available at https://github.com/Hsu1023/DuQuant-v2.

Yunfan Liu, Qi Li, Zhenan Sun et al.

One-shot face re-enactment is a challenging task due to the identity mismatch between source and driving faces. Specifically, the suboptimally disentangled identity information of driving subjects would inevitably interfere with the re-enactment results and lead to face shape distortion. To solve this problem, this paper proposes to use 3D Morphable Model (3DMM) for explicit facial semantic decomposition and identity disentanglement. Instead of using 3D coefficients alone for re-enactment control, we take the advantage of the generative ability of 3DMM to render textured face proxies. These proxies contain abundant yet compact geometric and semantic information of human faces, which enable us to compute the face motion field between source and driving images by estimating the dense correspondence. In this way, we could approximate re-enactment results by warping source images according to the motion field, and a Generative Adversarial Network (GAN) is adopted to further improve the visual quality of warping results. Extensive experiments on various datasets demonstrate the advantages of the proposed method over existing start-of-the-art benchmarks in both identity preservation and re-enactment fulfillment.

Jianxin Sun, Qiyao Deng, Qi Li et al.

Existing text-to-image synthesis methods generally are only applicable to words in the training dataset. However, human faces are so variable to be described with limited words. So this paper proposes the first free-style text-to-face method namely AnyFace enabling much wider open world applications such as metaverse, social media, cosmetics, forensics, etc. AnyFace has a novel two-stream framework for face image synthesis and manipulation given arbitrary descriptions of the human face. Specifically, one stream performs text-to-face generation and the other conducts face image reconstruction. Facial text and image features are extracted using the CLIP (Contrastive Language-Image Pre-training) encoders. And a collaborative Cross Modal Distillation (CMD) module is designed to align the linguistic and visual features across these two streams. Furthermore, a Diverse Triplet Loss (DT loss) is developed to model fine-grained features and improve facial diversity. Extensive experiments on Multi-modal CelebA-HQ and CelebAText-HQ demonstrate significant advantages of AnyFace over state-of-the-art methods. AnyFace can achieve high-quality, high-resolution, and high-diversity face synthesis and manipulation results without any constraints on the number and content of input captions.

Yunfan Liu, Qi Li, Qiyao Deng et al.

Facial Attribute Manipulation (FAM) aims to aesthetically modify a given face image to render desired attributes, which has received significant attention due to its broad practical applications ranging from digital entertainment to biometric forensics. In the last decade, with the remarkable success of Generative Adversarial Networks (GANs) in synthesizing realistic images, numerous GAN-based models have been proposed to solve FAM with various problem formulation approaches and guiding information representations. This paper presents a comprehensive survey of GAN-based FAM methods with a focus on summarizing their principal motivations and technical details. The main contents of this survey include: (i) an introduction to the research background and basic concepts related to FAM, (ii) a systematic review of GAN-based FAM methods in three main categories, and (iii) an in-depth discussion of important properties of FAM methods, open issues, and future research directions. This survey not only builds a good starting point for researchers new to this field but also serves as a reference for the vision community.

ShaoZhen Liu, Xinting Huang, Houwen Peng et al.

In recent years, large language models (LLMs) have demonstrated significant potential in complex reasoning tasks like mathematical problem-solving. However, existing research predominantly relies on reinforcement learning (RL) frameworks while overlooking supervised fine-tuning (SFT) methods. This paper proposes a new two-stage training framework that enhances models' self-correction capabilities through self-generated long chain-of-thought (CoT) data. During the first stage, a multi-turn dialogue strategy guides the model to generate CoT data incorporating verification, backtracking, subgoal decomposition, and backward reasoning, with predefined rules filtering high-quality samples for supervised fine-tuning. The second stage employs a difficulty-aware rejection sampling mechanism to dynamically optimize data distribution, strengthening the model's ability to handle complex problems. The approach generates reasoning chains extended over 4 times longer while maintaining strong scalability, proving that SFT effectively activates models' intrinsic reasoning capabilities and provides a resource-efficient pathway for complex task optimization. Experimental results demonstrate performance improvements on mathematical benchmarks including GSM8K and MATH500, with the fine-tuned model achieving a substantial improvement on competition-level problems like AIME24. Code will be open-sourced.

Xingqun Qi, Zhuojie Wu, Min Ren et al.

Efficient medical image segmentation aims to provide accurate pixel-wise predictions for medical images with a lightweight implementation framework. However, lightweight frameworks generally fail to achieve superior performance and suffer from poor generalizable ability on cross-domain tasks. In this paper, we explore the generalizable knowledge distillation for the efficient segmentation of cross-domain medical images. Considering the domain gaps between different medical datasets, we propose the Model-Specific Alignment Networks (MSAN) to obtain the domain-invariant representations. Meanwhile, a customized Alignment Consistency Training (ACT) strategy is designed to promote the MSAN training. Considering the domain-invariant representative vectors in MSAN, we propose two generalizable knowledge distillation schemes for cross-domain distillation, Dual Contrastive Graph Distillation (DCGD) and Domain-Invariant Cross Distillation (DICD). Specifically, in DCGD, two types of implicit contrastive graphs are designed to represent the intra-coupling and inter-coupling semantic correlations from the perspective of data distribution. In DICD, the domain-invariant semantic vectors from the two models (i.e., teacher and student) are leveraged to cross-reconstruct features by the header exchange of MSAN, which achieves improvement in the generalization of both the encoder and decoder in the student model. Furthermore, a metric named Frechet Semantic Distance (FSD) is tailored to verify the effectiveness of the regularized domain-invariant features. Extensive experiments conducted on the Liver and Retinal Vessel Segmentation datasets demonstrate the superiority of our method, in terms of performance and generalization on lightweight frameworks.

Lianwei Yang, Haisong Gong, Haokun Lin et al.

Vision Transformers (ViTs) have gained significant attention, but their high computing cost limits the practical applications. While post-training quantization (PTQ) reduces model size and speeds up inference, it often degrades performance, especially in low-bit settings. We identify two key reasons for the performance degradation: 1) existing quantization methods fail to align with the power-law distribution of post-Softmax activations, and 2) reparameterizing post-LayerNorm activations leads to a performance drop due to the significant influence of outliers in the scaling factors. To address these challenges, we propose DopQ-ViT, a Distribution-friendly and Outlier-aware Post-training Quantization method for ViTs. First, DopQ-ViT introduces the Tan Quantizer (TanQ), which better preserves the power-law distribution of post-Softmax activations by focusing more on values near 1. Second, DopQ-ViT presents the MAD-guided Optimal Scaling Factor (MOSF), which selects the optimal scaling factor without introducing additional calculations. Extensive experiments across various ViT models and quantization settings demonstrate that DopQ-ViT, with the help of TanQ and MOSF, outperforms previous PTQ methods on both classification and detection tasks.

Haokun Lin, Teng Wang, Yixiao Ge et al.

Pioneering token-based works such as Chameleon and Emu3 have established a foundation for multimodal unification but face challenges of high training computational overhead and limited comprehension performance due to a lack of high-level semantics. In this paper, we introduce TokLIP, a visual tokenizer that enhances comprehension by semanticizing vector-quantized (VQ) tokens and incorporating CLIP-level semantics while enabling end-to-end multimodal autoregressive training with standard VQ tokens. TokLIP integrates a low-level discrete VQ tokenizer with a ViT-based token encoder to capture high-level continuous semantics. Unlike previous approaches (e.g., VILA-U) that discretize high-level features, TokLIP disentangles training objectives for comprehension and generation, allowing the direct application of advanced VQ tokenizers without the need for tailored quantization operations. Our empirical results demonstrate that TokLIP achieves exceptional data efficiency, empowering visual tokens with high-level semantic understanding while enhancing low-level generative capacity, making it well-suited for autoregressive Transformers in both comprehension and generation tasks. The code and models are available at https://github.com/TencentARC/TokLIP.

Hang Zou, Chenxi Du, Ajian Liu et al.

Iris recognition is widely used in high-security scenarios due to its stability and distinctiveness. However, iris images captured by different devices exhibit certain and device-related consistent differences, which has a greater impact on the classification algorithm for anti-spoofing. The iris of various races would also affect the classification, causing the risk of identity theft. So it is necessary to improve the cross-domain capabilities of the iris anti-spoofing (IAS) methods to enable it more robust in facing different races and devices. However, there is no existing protocol that is comprehensively available. To address this gap, we propose an Iris Anti-Spoofing Cross-Domain-Testing (IAS-CDT) Protocol, which involves 10 datasets, belonging to 7 databases, published by 4 institutions, and collected with 6 different devices. It contains three sub-protocols hierarchically, aimed at evaluating average performance, cross-racial generalization, and cross-device generalization of IAS models. Moreover, to address the cross-device generalization challenge brought by the IAS-CDT Protocol, we employ multiple model parameter sets to learn from the multiple sub-datasets. Specifically, we utilize the Mixture of Experts (MoE) to fit complex data distributions using multiple sub-neural networks. To further enhance the generalization capabilities, we propose a novel method Masked-MoE (MMoE), which randomly masks a portion of tokens for some experts and requires their outputs to be similar to the unmasked experts, which can effectively mitigate the overfitting issue of MoE. For the evaluation, we selected ResNet50, VIT-B/16, CLIP, and FLIP as representative models and benchmarked them under the proposed IAS-CDT Protocol.

Hongda Liu, Yunfan Liu, Min Ren et al.

In skeleton-based action recognition, a key challenge is distinguishing between actions with similar trajectories of joints due to the lack of image-level details in skeletal representations. Recognizing that the differentiation of similar actions relies on subtle motion details in specific body parts, we direct our approach to focus on the fine-grained motion of local skeleton components. To this end, we introduce ProtoGCN, a Graph Convolutional Network (GCN)-based model that breaks down the dynamics of entire skeleton sequences into a combination of learnable prototypes representing core motion patterns of action units. By contrasting the reconstruction of prototypes, ProtoGCN can effectively identify and enhance the discriminative representation of similar actions. Without bells and whistles, ProtoGCN achieves state-of-the-art performance on multiple benchmark datasets, including NTU RGB+D, NTU RGB+D 120, Kinetics-Skeleton, and FineGYM, which demonstrates the effectiveness of the proposed method. The code is available at https://github.com/firework8/ProtoGCN.

Haokun Lin, Haobo Xu, Yichen Wu et al.

Recent advances in diffusion large language models (dLLMs) have introduced a promising alternative to autoregressive (AR) LLMs for natural language generation tasks, leveraging full attention and denoising-based decoding strategies. However, the deployment of these models on edge devices remains challenging due to their massive parameter scale and high resource demands. While post-training quantization (PTQ) has emerged as a widely adopted technique for compressing AR LLMs, its applicability to dLLMs remains largely unexplored. In this work, we present the first systematic study on quantizing diffusion-based language models. We begin by identifying the presence of activation outliers, characterized by abnormally large activation values that dominate the dynamic range. These outliers pose a key challenge to low-bit quantization, as they make it difficult to preserve precision for the majority of values. More importantly, we implement state-of-the-art PTQ methods and conduct a comprehensive evaluation across multiple task types and model variants. Our analysis is structured along four key dimensions: bit-width, quantization method, task category, and model type. Through this multi-perspective evaluation, we offer practical insights into the quantization behavior of dLLMs under different configurations. We hope our findings provide a foundation for future research in efficient dLLM deployment. Our code is publicly available at https://github.com/FelixMessi/QDLM.

Mengmeng Cui, Kunbo Zhang, Zhenan Sun

In recent years, 2D-to-3D pose uplifting in monocular 3D Human Pose Estimation (HPE) has attracted widespread research interest. GNN-based methods and Transformer-based methods have become mainstream architectures due to their advanced spatial and temporal feature learning capacities. However, existing approaches typically construct joint-wise and frame-wise attention alignments in spatial and temporal domains, resulting in dense connections that introduce considerable local redundancy and computational overhead. In this paper, we take a global approach to exploit spatio-temporal information and realise efficient 3D HPE with a concise Graph and Skipped Transformer architecture. Specifically, in Spatial Encoding stage, coarse-grained body parts are deployed to construct Spatial Graph Network with a fully data-driven adaptive topology, ensuring model flexibility and generalizability across various poses. In Temporal Encoding and Decoding stages, a simple yet effective Skipped Transformer is proposed to capture long-range temporal dependencies and implement hierarchical feature aggregation. A straightforward Data Rolling strategy is also developed to introduce dynamic information into 2D pose sequence. Extensive experiments are conducted on Human3.6M, MPI-INF-3DHP and Human-Eva benchmarks. G-SFormer series methods achieve superior performances compared with previous state-of-the-arts with only around ten percent of parameters and significantly reduced computational complexity. Additionally, G-SFormer also exhibits outstanding robustness to inaccuracies in detected 2D poses.

Zhiwei Li, Jiacheng Xue, Weining Wang et al.

Vision-language models like CLIP have demonstrated remarkable zero-shot transfer capabilities. However, their susceptibility to imperceptible adversarial perturbations remains a critical security concern. While test-time defenses offer a pragmatic solution for deployed models, existing approaches typically rely on gradient-based optimization during inference, incurring significant computational overhead. In this paper, we revisit the role of data augmentation in CLIP robustness and observe that augmentations are not equally effective: specific augmentations consistently provide robust geometric cues that align with correct class semantics in the hyperspherical feature space. Based on this, we propose Adaptive Geodesic Correction (AGC), a training-free defense mechanism that requires no parameter updates. AGC identifies a reliable augmentation as a geometric anchor and corrects the input feature towards it, utilizing an adaptive step size to balance robustness against clean accuracy preservation. AGC achieves superior performance across eight fine-grained datasets and three CLIP backbones, improving average robust accuracy by 44.4\% over state-of-the-art baseline while delivering a 10$\times$ reduction in inference latency. Our findings reveal a fundamental geometric property of CLIP features, offering a highly efficient and effective paradigm for robust multimodal deployment.

Muyi Sun, Yixuan Wang, Hong Wang et al.

Audio-Visual Learning (AVL) is one fundamental task of multi-modality learning and embodied intelligence, displaying the vital role in scene understanding and interaction. However, previous researchers mostly focus on exploring downstream tasks from a coarse-grained perspective (e.g., audio-visual correspondence, sound source localization, and audio-visual event localization). Considering providing more specific scene perception details, we newly define a fine-grained Audio-Visual Learning task, termed Region-Aware Sound Source Understanding (RA-SSU), which aims to achieve region-aware, frame-level, and high-quality sound source understanding. To support this goal, we innovatively construct two corresponding datasets, i.e. fine-grained Music (f-Music) and fine-grained Lifescene (f-Lifescene), each containing annotated sound source masks and frame-by-frame textual descriptions. The f-Music dataset includes 3,976 samples across 22 scene types related to specific application scenarios, focusing on music scenes with complex instrument mixing. The f-Lifescene dataset contains 6,156 samples across 61 types representing diverse sounding objects in life scenarios. Moreover, we propose SSUFormer, a Sound-Source Understanding TransFormer benchmark that facilitates both the sound source segmentation and sound region description with a multi-modal input and multi-modal output architecture. Specifically, we design two modules for this framework, Mask Collaboration Module (MCM) and Mixture of Hierarchical-prompted Experts (MoHE), to respectively enhance the accuracy and enrich the elaboration of the sound source description. Extensive experiments are conducted on our two datasets to verify the feasibility of the task, evaluate the availability of the datasets, and demonstrate the superiority of the SSUFormer, which achieves SOTA performance on the Sound Source Understanding benchmark.

Zhiwei Li, Yitian Pang, Weining Wang et al.

Vision-Language Models (VLMs), such as CLIP, have achieved impressive zero-shot recognition performance but remain highly susceptible to adversarial perturbations, posing significant risks in safety-critical scenarios. Previous training-time defenses rely on adversarial fine-tuning, which requires labeled data and costly retraining, while existing test-time strategies fail to reliably distinguish between clean and adversarial inputs, thereby preventing both adversarial robustness and clean accuracy from reaching their optimum. To address these limitations, we propose Test-Time Padding (TTP), a lightweight defense framework that performs adversarial detection followed by targeted adaptation at inference. TTP identifies adversarial inputs via the cosine similarity shift between CLIP feature embeddings computed before and after spatial padding, yielding a universal threshold for reliable detection across architectures and datasets. For detected adversarial cases, TTP employs trainable padding to restore disrupted attention patterns, coupled with a similarity-aware ensemble strategy for a more robust final prediction. For clean inputs, TTP leaves them unchanged by default or optionally integrates existing test-time adaptation techniques for further accuracy gains. Comprehensive experiments on diverse CLIP backbones and fine-grained benchmarks show that TTP consistently surpasses state-of-the-art test-time defenses, delivering substantial improvements in adversarial robustness without compromising clean accuracy. The code for this paper will be released soon.

Xinyang Song, Libin Wang, Weining Wang et al.

Recent image generation approaches often address subject, style, and structure-driven conditioning in isolation, leading to feature entanglement and limited task transferability. In this paper, we introduce 3SGen, a task-aware unified framework that performs all three conditioning modes within a single model. 3SGen employs an MLLM equipped with learnable semantic queries to align text-image semantics, complemented by a VAE branch that preserves fine-grained visual details. At its core, an Adaptive Task-specific Memory (ATM) module dynamically disentangles, stores, and retrieves condition-specific priors, such as identity for subjects, textures for styles, and spatial layouts for structures, via a lightweight gating mechanism along with several scalable memory items. This design mitigates inter-task interference and naturally scales to compositional inputs. In addition, we propose 3SGen-Bench, a unified image-driven generation benchmark with standardized metrics for evaluating cross-task fidelity and controllability. Extensive experiments on our proposed 3SGen-Bench and other public benchmarks demonstrate our superior performance across diverse image-driven generation tasks.

Hongda Liu, Yunfan Liu, Changlu Wang et al.

Recent advances in skeleton-based action recognition increasingly leverage semantic priors from Large Language Models (LLMs) to enrich skeletal representations. However, the LLM is typically queried in isolation from the recognition model and receives no performance feedback. As a result, it often fails to deliver the targeted discriminative cues critical to distinguish similar actions. To overcome these limitations, we propose SkeletonAgent, a novel framework that bridges the recognition model and the LLM through two cooperative agents, i.e., Questioner and Selector. Specifically, the Questioner identifies the most frequently confused classes and supplies them to the LLM as context for more targeted guidance. Conversely, the Selector parses the LLM's response to extract precise joint-level constraints and feeds them back to the recognizer, enabling finer-grained cross-modal alignment. Comprehensive evaluations on five benchmarks, including NTU RGB+D, NTU RGB+D 120, Kinetics-Skeleton, FineGYM, and UAV-Human, demonstrate that SkeletonAgent consistently outperforms state-of-the-art benchmark methods. The code is available at https://github.com/firework8/SkeletonAgent.

Shujun Xia, Haokun Lin, Yichen Wu et al.

LLMs hold great promise for healthcare applications, but the rapid evolution of medical knowledge and errors in training data often cause them to generate outdated or inaccurate information, limiting their applicability in high-stakes clinical practice. Model editing has emerged as a potential remedy without full retraining. While parameter-based editing often compromises locality and is thus ill-suited for the medical domain, retrieval-based editing offers a more viable alternative. However, it still faces two critical challenges: (1) representation overlap within the medical knowledge space often causes inaccurate retrieval and reduces editing accuracy; (2) existing methods are restricted to single-sample edits, while batch-editing remains largely unexplored despite its importance for real-world medical applications. To address these challenges, we first construct MedVersa, an enhanced benchmark with broader coverage of medical subjects, designed to evaluate both single and batch edits under strict locality constraints. We then propose MedREK, a retrieval-based editing framework that integrates a shared query-key module for precise matching with an attention-based prompt encoder for informative guidance. Experimental results on various medical benchmarks demonstrate that our MedREK achieves superior performance across different core metrics and provides the first validated solution for batch-editing in medical LLMs. Our code and dataset are available at https://github.com/mylittleriver/MedREK.

Zixuan Li, Binzong Geng, Jing Xiong et al.

Click-Through Rate (CTR) prediction, a core task in recommendation systems, estimates user click likelihood using historical behavioral data. Modeling user behavior sequences as text to leverage Language Models (LMs) for this task has gained traction, owing to LMs' strong semantic understanding and contextual modeling capabilities. However, a critical structural gap exists: user behavior sequences consist of discrete actions connected by semantically empty separators, differing fundamentally from the coherent natural language in LM pre-training. This mismatch causes semantic fragmentation, where LM attention scatters across irrelevant tokens instead of focusing on meaningful behavior boundaries and inter-behavior relationships, degrading prediction performance. To address this, we propose $\textit{CTR-Sink}$, a novel framework introducing behavior-level attention sinks tailored for recommendation scenarios. Inspired by attention sink theory, it constructs attention focus sinks and dynamically regulates attention aggregation via external information. Specifically, we insert sink tokens between consecutive behaviors, incorporating recommendation-specific signals such as temporal distance to serve as stable attention sinks. To enhance generality, we design a two-stage training strategy that explicitly guides LM attention toward sink tokens and a attention sink mechanism that amplifies inter-sink dependencies to better capture behavioral correlations. Experiments on one industrial dataset and two open-source datasets (MovieLens, Kuairec), alongside visualization results, validate the method's effectiveness across scenarios.

Yinfan Wang, Jie Gui, Baosheng Yu et al.

A major challenge in finger vein recognition is the lack of large-scale public datasets. Existing datasets contain few identities and limited samples per finger, restricting the advancement of deep learning-based methods. To address this, we introduce FVeinSyn, a synthetic generator capable of producing diverse finger vein patterns with rich intra-class variations. Using FVeinSyn, we created FingerVeinSyn-5M -- the largest available finger vein dataset -- containing 5 million samples from 50,000 unique fingers, each with 100 variations including shift, rotation, scale, roll, varying exposure levels, skin scattering blur, optical blur, and motion blur. FingerVeinSyn-5M is also the first to offer fully annotated finger vein images, supporting deep learning applications in this field. Models pretrained on FingerVeinSyn-5M and fine-tuned with minimal real data achieve an average 53.91\% performance gain across multiple benchmarks. The dataset is publicly available at: https://github.com/EvanWang98/FingerVeinSyn-5M.

Haokun Lin, Haobo Xu, Yichen Wu et al.

Quantization of large language models (LLMs) faces significant challenges, particularly due to the presence of outlier activations that impede efficient low-bit representation. Traditional approaches predominantly address Normal Outliers, which are activations across all tokens with relatively large magnitudes. However, these methods struggle with smoothing Massive Outliers that display significantly larger values, which leads to significant performance degradation in low-bit quantization. In this paper, we introduce DuQuant, a novel approach that utilizes rotation and permutation transformations to more effectively mitigate both massive and normal outliers. First, DuQuant starts by constructing the rotation matrix, using specific outlier dimensions as prior knowledge, to redistribute outliers to adjacent channels by block-wise rotation. Second, We further employ a zigzag permutation to balance the distribution of outliers across blocks, thereby reducing block-wise variance. A subsequent rotation further smooths the activation landscape, enhancing model performance. DuQuant simplifies the quantization process and excels in managing outliers, outperforming the state-of-the-art baselines across various sizes and types of LLMs on multiple tasks, even with 4-bit weight-activation quantization. Our code is available at https://github.com/Hsu1023/DuQuant.

Boqiang Xu, Jian Liang, Lingxiao He et al.

Domain generalizable (DG) person re-identification (ReID) aims to test across unseen domains without access to the target domain data at training time, which is a realistic but challenging problem. In contrast to methods assuming an identical model for different domains, Mixture of Experts (MoE) exploits multiple domain-specific networks for leveraging complementary information between domains, obtaining impressive results. However, prior MoE-based DG ReID methods suffer from a large model size with the increase of the number of source domains, and most of them overlook the exploitation of domain-invariant characteristics. To handle the two issues above, this paper presents a new approach called Mimic Embedding via adapTive Aggregation (META) for DG person ReID. To avoid the large model size, experts in META do not adopt a branch network for each source domain but share all the parameters except for the batch normalization layers. Besides multiple experts, META leverages Instance Normalization (IN) and introduces it into a global branch to pursue invariant features across domains. Meanwhile, META considers the relevance of an unseen target sample and source domains via normalization statistics and develops an aggregation module to adaptively integrate multiple experts for mimicking unseen target domain. Benefiting from a proposed consistency loss and an episodic training algorithm, META is expected to mimic embedding for a truly unseen target domain. Extensive experiments verify that META surpasses state-of-the-art DG person ReID methods by a large margin. Our code is available at https://github.com/xbq1994/META.

Leyuan Wang, Kunbo Zhang, Yunlong Wang et al.

For distant iris recognition, a long focal length lens is generally used to ensure the resolution ofiris images, which reduces the depth of field and leads to potential defocus blur. To accommodate users at different distances, it is necessary to control focus quickly and accurately. While for users in motion, it is expected to maintain the correct focus on the iris area continuously. In this paper, we introduced a novel rapid autofocus camera for active refocusing ofthe iris area ofthe moving objects using a focus-tunable lens. Our end-to-end computational algorithm can predict the best focus position from one single blurred image and generate a lens diopter control signal automatically. This scene-based active manipulation method enables real-time focus tracking of the iris area ofa moving object. We built a testing bench to collect real-world focal stacks for evaluation of the autofocus methods. Our camera has reached an autofocus speed ofover 50 fps. The results demonstrate the advantages of our proposed camera for biometric perception in static and dynamic scenes. The code is available at https://github.com/Debatrix/AquulaCam.

Leyuan Wang, Kunbo Zhang, Min Ren et al.

A large portion of iris images captured in real world scenarios are poor quality due to the uncontrolled environment and the non-cooperative subject. To ensure that the recognition algorithm is not affected by low-quality images, traditional hand-crafted factors based methods discard most images, which will cause system timeout and disrupt user experience. In this paper, we propose a recognition-oriented quality metric and assessment method for iris image to deal with the problem. The method regards the iris image embeddings Distance in Feature Space (DFS) as the quality metric and the prediction is based on deep neural networks with the attention mechanism. The quality metric proposed in this paper can significantly improve the performance of the recognition algorithm while reducing the number of images discarded for recognition, which is advantageous over hand-crafted factors based iris quality assessment methods. The relationship between Image Rejection Rate (IRR) and Equal Error Rate (EER) is proposed to evaluate the performance of the quality assessment algorithm under the same image quality distribution and the same recognition algorithm. Compared with hand-crafted factors based methods, the proposed method is a trial to bridge the gap between the image quality assessment and biometric recognition. The code is available at https://github.com/Debatrix/DFSNet.

Boqiang Xu, Lingxiao He, Xingyu Liao et al.

Person re-identification (Re-ID) aims at retrieving an input person image from a set of images captured by multiple cameras. Although recent Re-ID methods have made great success, most of them extract features in terms of the attributes of clothing (e.g., color, texture). However, it is common for people to wear black clothes or be captured by surveillance systems in low light illumination, in which cases the attributes of the clothing are severely missing. We call this problem the Black Re-ID problem. To solve this problem, rather than relying on the clothing information, we propose to exploit head-shoulder features to assist person Re-ID. The head-shoulder adaptive attention network (HAA) is proposed to learn the head-shoulder feature and an innovative ensemble method is designed to enhance the generalization of our model. Given the input person image, the ensemble method would focus on the head-shoulder feature by assigning a larger weight if the individual insides the image is in black clothing. Due to the lack of a suitable benchmark dataset for studying the Black Re-ID problem, we also contribute the first Black-reID dataset, which contains 1274 identities in training set. Extensive evaluations on the Black-reID, Market1501 and DukeMTMC-reID datasets show that our model achieves the best result compared with the state-of-the-art Re-ID methods on both Black and conventional Re-ID problems. Furthermore, our method is also proved to be effective in dealing with person Re-ID in similar clothing. Our code and dataset are avaliable on https://github.com/xbq1994/.

Min Ren, Yunlong Wang, Zhenan Sun et al.

The generalization ability of Convolutional neural networks (CNNs) for biometrics drops greatly due to the adverse effects of various occlusions. To this end, we propose a novel unified framework integrated the merits of both CNNs and graphical models to learn dynamic graph representations for occlusion problems in biometrics, called Dynamic Graph Representation (DGR). Convolutional features onto certain regions are re-crafted by a graph generator to establish the connections among the spatial parts of biometrics and build Feature Graphs based on these node representations. Each node of Feature Graphs corresponds to a specific part of the input image and the edges express the spatial relationships between parts. By analyzing the similarities between the nodes, the framework is able to adaptively remove the nodes representing the occluded parts. During dynamic graph matching, we propose a novel strategy to measure the distances of both nodes and adjacent matrixes. In this way, the proposed method is more convincing than CNNs-based methods because the dynamic graph method implies a more illustrative and reasonable inference of the biometrics decision. Experiments conducted on iris and face demonstrate the superiority of the proposed framework, which boosts the accuracy of occluded biometrics recognition by a large margin comparing with baseline methods.The code is avaliable at https://github.com/RenMin1991/Dyamic\_Graph\_Representation

Caiyong Wang, Yuhao Zhu, Yunfan Liu et al.

Iris segmentation and localization in non-cooperative environment is challenging due to illumination variations, long distances, moving subjects and limited user cooperation, etc. Traditional methods often suffer from poor performance when confronted with iris images captured in these conditions. Recent studies have shown that deep learning methods could achieve impressive performance on iris segmentation task. In addition, as iris is defined as an annular region between pupil and sclera, geometric constraints could be imposed to help locating the iris more accurately and improve the segmentation results. In this paper, we propose a deep multi-task learning framework, named as IrisParseNet, to exploit the inherent correlations between pupil, iris and sclera to boost up the performance of iris segmentation and localization in a unified model. In particular, IrisParseNet firstly applies a Fully Convolutional Encoder-Decoder Attention Network to simultaneously estimate pupil center, iris segmentation mask and iris inner/outer boundary. Then, an effective post-processing method is adopted for iris inner/outer circle localization.To train and evaluate the proposed method, we manually label three challenging iris datasets, namely CASIA-Iris-Distance, UBIRIS.v2, and MICHE-I, which cover various types of noises. Extensive experiments are conducted on these newly annotated datasets, and results show that our method outperforms state-of-the-art methods on various benchmarks. All the ground-truth annotations, annotation codes and evaluation protocols are publicly available at https://github.com/xiamenwcy/IrisParseNet.

Lingxiao He, Zhenan Sun, Yuhao Zhu et al.

Biometric recognition on partial captured targets is challenging, where only several partial observations of objects are available for matching. In this area, deep learning based methods are widely applied to match these partial captured objects caused by occlusions, variations of postures or just partial out of view in person re-identification and partial face recognition. However, most current methods are not able to identify an individual in case that some parts of the object are not obtainable, while the rest are specialized to certain constrained scenarios. To this end, we propose a robust general framework for arbitrary biometric matching scenarios without the limitations of alignment as well as the size of inputs. We introduce a feature post-processing step to handle the feature maps from FCN and a dictionary learning based Spatial Feature Reconstruction (SFR) to match different sized feature maps in this work. Moreover, the batch hard triplet loss function is applied to optimize the model. The applicability and effectiveness of the proposed method are demonstrated by the results from experiments on three person re-identification datasets (Market1501, CUHK03, DukeMTMC-reID), two partial person datasets (Partial REID and Partial iLIDS) and two partial face datasets (CASIA-NIR-Distance and Partial LFW), on which state-of-the-art performance is ensured in comparison with several state-of-the-art approaches. The code is released online and can be found on the website: https://github.com/lingxiao-he/Partial-Person-ReID.

Xiang Wu, Ran He, Zhenan Sun et al.

The volume of convolutional neural network (CNN) models proposed for face recognition has been continuously growing larger to better fit large amount of training data. When training data are obtained from internet, the labels are likely to be ambiguous and inaccurate. This paper presents a Light CNN framework to learn a compact embedding on the large-scale face data with massive noisy labels. First, we introduce a variation of maxout activation, called Max-Feature-Map (MFM), into each convolutional layer of CNN. Different from maxout activation that uses many feature maps to linearly approximate an arbitrary convex activation function, MFM does so via a competitive relationship. MFM can not only separate noisy and informative signals but also play the role of feature selection between two feature maps. Second, three networks are carefully designed to obtain better performance meanwhile reducing the number of parameters and computational costs. Lastly, a semantic bootstrapping method is proposed to make the prediction of the networks more consistent with noisy labels. Experimental results show that the proposed framework can utilize large-scale noisy data to learn a Light model that is efficient in computational costs and storage spaces. The learned single network with a 256-D representation achieves state-of-the-art results on various face benchmarks without fine-tuning. The code is released on https://github.com/AlfredXiangWu/LightCNN.

Haokun Lin, Haoli Bai, Zhili Liu et al.

Vision-language pre-trained models have achieved impressive performance on various downstream tasks. However, their large model sizes hinder their utilization on platforms with limited computational resources. We find that directly using smaller pre-trained models and applying magnitude-based pruning on CLIP models leads to inflexibility and inferior performance. Recent efforts for VLP compression either adopt uni-modal compression metrics resulting in limited performance or involve costly mask-search processes with learnable masks. In this paper, we first propose the Module-wise Pruning Error (MoPE) metric, accurately assessing CLIP module importance by performance decline on cross-modal tasks. Using the MoPE metric, we introduce a unified pruning framework applicable to both pre-training and task-specific fine-tuning compression stages. For pre-training, MoPE-CLIP effectively leverages knowledge from the teacher model, significantly reducing pre-training costs while maintaining strong zero-shot capabilities. For fine-tuning, consecutive pruning from width to depth yields highly competitive task-specific models. Extensive experiments in two stages demonstrate the effectiveness of the MoPE metric, and MoPE-CLIP outperforms previous state-of-the-art VLP compression methods.

Beiyuan Zhang, Yue Ma, Chunlei Fu et al.

Text-editable and pose-controllable character video generation is a challenging but prevailing topic with practical applications. However, existing approaches mainly focus on single-object video generation with pose guidance, ignoring the realistic situation that multi-character appear concurrently in a scenario. To tackle this, we propose a novel multi-character video generation framework in a tuning-free manner, which is based on the separated text and pose guidance. Specifically, we first extract character masks from the pose sequence to identify the spatial position for each generating character, and then single prompts for each character are obtained with LLMs for precise text guidance. Moreover, the spatial-aligned cross attention and multi-branch control module are proposed to generate fine grained controllable multi-character video. The visualized results of generating video demonstrate the precise controllability of our method for multi-character generation. We also verify the generality of our method by applying it to various personalized T2I models. Moreover, the quantitative results show that our approach achieves superior performance compared with previous works.

Zhihan Ju, Wanting Zhou, Longteng Kong et al.

Medical Image Synthesis (MIS) plays an important role in the intelligent medical field, which greatly saves the economic and time costs of medical diagnosis. However, due to the complexity of medical images and similar characteristics of different tissue cells, existing methods face great challenges in meeting their biological consistency. To this end, we propose the Hybrid Augmented Generative Adversarial Network (HAGAN) to maintain the authenticity of structural texture and tissue cells. HAGAN contains Attention Mixed (AttnMix) Generator, Hierarchical Discriminator and Reverse Skip Connection between Discriminator and Generator. The AttnMix consistency differentiable regularization encourages the perception in structural and textural variations between real and fake images, which improves the pathological integrity of synthetic images and the accuracy of features in local areas. The Hierarchical Discriminator introduces pixel-by-pixel discriminant feedback to generator for enhancing the saliency and discriminance of global and local details simultaneously. The Reverse Skip Connection further improves the accuracy for fine details by fusing real and synthetic distribution features. Our experimental evaluations on three datasets of different scales, i.e., COVID-CT, ACDC and BraTS2018, demonstrate that HAGAN outperforms the existing methods and achieves state-of-the-art performance in both high-resolution and low-resolution.

Fangling Jiang, Qi Li, Weining Wang et al.

Recently the emergence of novel presentation attacks has drawn increasing attention to face anti-spoofing. However, existing methods tend to memorize data patterns from the training set, resulting in poor generalization to unknown attack types across different scenarios and limited interpretability. To address these challenges, this paper presents a reinforcement fine-tuning-based face anti-spoofing method that stimulates the capabilities of multimodal large language models to think and learn how to solve the anti-spoofing task itself, rather than relying on the memorization of authenticity patterns. We design verifiable class consistent reward and reasoning consistent reward, and employ a GRPO-based optimization strategy to guide the model in exploring reasoning policies from multiple perspectives to maximize expected rewards. As a result, through iterative trial-and-error learning while retaining only high-reward trajectories, the model distills highly generalizable decision-making rules from the extensive solution space to effectively address cross-domain face anti-spoofing tasks. Extensive experimental results demonstrate that our method achieves state-of-the-art cross-domain generalization performance. It generalizes well to diverse unknown attack types in unseen target domains while providing interpretable reasoning for its authenticity decisions without requiring labor-intensive textual annotations for training.

Xinyang Song, Libin Wang, Weining Wang et al.

The remarkable success of diffusion models in text-to-image generation has sparked growing interest in expanding their capabilities to a variety of multi-modal tasks, including image understanding, manipulation, and perception. These tasks require advanced semantic comprehension across both visual and textual modalities, especially in scenarios involving complex semantic instructions. However, existing approaches often rely heavily on vision-language models (VLMs) or modular designs for semantic guidance, leading to fragmented architectures and computational inefficiency. To address these challenges, we propose UniAlignment, a unified multimodal generation framework within a single diffusion transformer. UniAlignment introduces a dual-stream diffusion training strategy that incorporates both intrinsic-modal semantic alignment and cross-modal semantic alignment, thereby enhancing the model's cross-modal consistency and instruction-following robustness. Additionally, we present SemGen-Bench, a new benchmark specifically designed to evaluate multimodal semantic consistency under complex textual instructions. Extensive experiments across multiple tasks and benchmarks demonstrate that UniAlignment outperforms existing baselines, underscoring the significant potential of diffusion models in unified multimodal generation.

Lianwei Yang, Haokun Lin, Tianchen Zhao et al. · tsinghua

Diffusion Transformers (DiTs) have achieved impressive performance in text-to-image and text-to-video generation. However, their high computational cost and large parameter sizes pose significant challenges for usage in resource-constrained scenarios. Effective compression of models has become a crucial issue that urgently needs to be addressed. Post-training quantization (PTQ) is a promising solution to reduce memory usage and accelerate inference, but existing PTQ methods suffer from severe performance degradation under extreme low-bit settings. After experiments and analysis, we identify two key obstacles to low-bit PTQ for DiTs: (1) the weights of DiT models follow a Gaussian-like distribution with long tails, causing uniform quantization to poorly allocate intervals and leading to significant quantization errors. This issue has been observed in the linear layer weights of different DiT models, which deeply limits the performance. (2) two types of activation outliers in DiT models: (i) Mild Outliers with slightly elevated values, and (ii) Salient Outliers with large magnitudes concentrated in specific channels, which disrupt activation quantization. To address these issues, we propose LRQ-DiT, an efficient and accurate post-training quantization framework for image and video generation. First, we introduce Twin-Log Quantization (TLQ), a log-based method that allocates more quantization intervals to the intermediate dense regions, effectively achieving alignment with the weight distribution and reducing quantization errors. Second, we propose an Adaptive Rotation Scheme (ARS) that dynamically applies Hadamard or outlier-aware rotations based on activation fluctuation, effectively mitigating the impact of both types of outliers. Extensive experiments on various text-to-image and text-to-video DiT models demonstrate that LRQ-DiT preserves high generation quality.

Yizhi Hu, Zezhao Tian, Xingqun Qi et al.

Referring Expression Comprehension (REC) aims to localize specified entities or regions in an image based on natural language descriptions. While existing methods handle single-entity localization, they often ignore complex inter-entity relationships in multi-entity scenes, limiting their accuracy and reliability. Additionally, the lack of high-quality datasets with fine-grained, paired image-text-relation annotations hinders further progress. To address this challenge, we first construct a relation-aware, multi-entity REC dataset called ReMeX, which includes detailed relationship and textual annotations. We then propose ReMeREC, a novel framework that jointly leverages visual and textual cues to localize multiple entities while modeling their inter-relations. To address the semantic ambiguity caused by implicit entity boundaries in language, we introduce the Text-adaptive Multi-entity Perceptron (TMP), which dynamically infers both the quantity and span of entities from fine-grained textual cues, producing distinctive representations. Additionally, our Entity Inter-relationship Reasoner (EIR) enhances relational reasoning and global scene understanding. To further improve language comprehension for fine-grained prompts, we also construct a small-scale auxiliary dataset, EntityText, generated using large language models. Experiments on four benchmark datasets show that ReMeREC achieves state-of-the-art performance in multi-entity grounding and relation prediction, outperforming existing approaches by a large margin.