Zechao Li

Yanpeng Sun, Qiang Chen, Xiangyu He et al.

Freezing the pre-trained backbone has become a standard paradigm to avoid overfitting in few-shot segmentation. In this paper, we rethink the paradigm and explore a new regime: {\em fine-tuning a small part of parameters in the backbone}. We present a solution to overcome the overfitting problem, leading to better model generalization on learning novel classes. Our method decomposes backbone parameters into three successive matrices via the Singular Value Decomposition (SVD), then {\em only fine-tunes the singular values} and keeps others frozen. The above design allows the model to adjust feature representations on novel classes while maintaining semantic clues within the pre-trained backbone. We evaluate our {\em Singular Value Fine-tuning (SVF)} approach on various few-shot segmentation methods with different backbones. We achieve state-of-the-art results on both Pascal-5$^i$ and COCO-20$^i$ across 1-shot and 5-shot settings. Hopefully, this simple baseline will encourage researchers to rethink the role of backbone fine-tuning in few-shot settings. The source code and models will be available at https://github.com/syp2ysy/SVF.

Yanpeng Sun, Qiang Chen, Xiaofan Li et al.

The In-Context Learning (ICL) is to understand a new task via a few demonstrations (aka. prompt) and predict new inputs without tuning the models. While it has been widely studied in NLP, it is still a relatively new area of research in computer vision. To reveal the factors influencing the performance of visual in-context learning, this paper shows that prompt selection and prompt fusion are two major factors that have a direct impact on the inference performance of visual context learning. Prompt selection is the process of identifying the most appropriate prompt or example to help the model understand new tasks. This is important because providing the model with relevant prompts can help it learn more effectively and efficiently. Prompt fusion involves combining knowledge from different positions within the large-scale visual model. By doing this, the model can leverage the diverse knowledge stored in different parts of the model to improve its performance on new tasks. Based these findings, we propose a simple framework prompt-SelF for visual in-context learning. Specifically, we first use the pixel-level retrieval method to select a suitable prompt, and then use different prompt fusion methods to activate all the knowledge stored in the large-scale model, and finally ensemble the prediction results obtained from different prompt fusion methods to obtain the final prediction results. And we conduct extensive experiments on single-object segmentation and detection tasks to demonstrate the effectiveness of prompt-SelF. Remarkably, the prompt-SelF has outperformed OSLSM based meta-learning in 1-shot segmentation for the first time. This indicated the great potential of visual in-context learning. The source code and models will be available at https://github.com/syp2ysy/prompt-SelF.

Jinnan Yang, Yan Wang, Zhen Bi et al.

Diffusion Large Language Models (DLMs) have demonstrated significant advantages across various tasks. However, constrained by their multi-step iterative inference mechanism, their computational overhead and inference latency in long-context tasks have become core bottlenecks restricting their large-scale deployment. When processing long sequences, existing Key-Value (KV) caching mechanisms often face a dilemma where generation quality degrades drastically, where the core challenge lies in precisely and efficiently filtering critical tokens within ultra-long contexts. Inspired by the human reading process, we propose \textbf{WaveFilter}, a universal and training-free caching framework. This framework innovatively introduces the wavelet transform for decomposition of long sequences to achieve precise identification of key tokens, based on which a sparse KV Cache is constructed to compute the final contextual representation. Experimental results demonstrate that WaveFilter, as a plug-and-play generic framework, significantly enhances the performance of existing mainstream KV Cache methods in complex long-context tasks.

Xuejing Liu, Liang Li, Shuhui Wang et al.

Weakly supervised Referring Expression Grounding (REG) aims to ground a particular target in an image described by a language expression while lacking the correspondence between target and expression. Two main problems exist in weakly supervised REG. First, the lack of region-level annotations introduces ambiguities between proposals and queries. Second, most previous weakly supervised REG methods ignore the discriminative location and context of the referent, causing difficulties in distinguishing the target from other same-category objects. To address the above challenges, we design an entity-enhanced adaptive reconstruction network (EARN). Specifically, EARN includes three modules: entity enhancement, adaptive grounding, and collaborative reconstruction. In entity enhancement, we calculate semantic similarity as supervision to select the candidate proposals. Adaptive grounding calculates the ranking score of candidate proposals upon subject, location and context with hierarchical attention. Collaborative reconstruction measures the ranking result from three perspectives: adaptive reconstruction, language reconstruction and attribute classification. The adaptive mechanism helps to alleviate the variance of different referring expressions. Experiments on five datasets show EARN outperforms existing state-of-the-art methods. Qualitative results demonstrate that the proposed EARN can better handle the situation where multiple objects of a particular category are situated together.

Fei Shen, Xin Jiang, Xin He et al.

Latest advances have achieved realistic virtual try-on (VTON) through localized garment inpainting using latent diffusion models, significantly enhancing consumers' online shopping experience. However, existing VTON technologies neglect the need for merchants to showcase garments comprehensively, including flexible control over garments, optional faces, poses, and scenes. To address this issue, we define a virtual dressing (VD) task focused on generating freely editable human images with fixed garments and optional conditions. Meanwhile, we design a comprehensive affinity metric index (CAMI) to evaluate the consistency between generated images and reference garments. Then, we propose IMAGDressing-v1, which incorporates a garment UNet that captures semantic features from CLIP and texture features from VAE. We present a hybrid attention module, including a frozen self-attention and a trainable cross-attention, to integrate garment features from the garment UNet into a frozen denoising UNet, ensuring users can control different scenes through text. IMAGDressing-v1 can be combined with other extension plugins, such as ControlNet and IP-Adapter, to enhance the diversity and controllability of generated images. Furthermore, to address the lack of data, we release the interactive garment pairing (IGPair) dataset, containing over 300,000 pairs of clothing and dressed images, and establish a standard pipeline for data assembly. Extensive experiments demonstrate that our IMAGDressing-v1 achieves state-of-the-art human image synthesis performance under various controlled conditions. The code and model will be available at https://github.com/muzishen/IMAGDressing.

Shiyu Xuan, Dongkai Wang, Zechao Li et al.

Zero-shot Human-object interaction (HOI) detection aims to locate humans and objects in images and recognize their interactions. While advances in open-vocabulary object detection provide promising solutions for object localization, interaction recognition (IR) remains challenging due to the combinatorial diversity of interactions. Existing methods, including two-stage methods, tightly couple IR with a specific detector and rely on coarse-grained vision-language model (VLM) features, which limit generalization to unseen interactions. In this work, we propose a decoupled framework that separates object detection from IR and leverages multi-modal large language models (MLLMs) for zero-shot IR. We introduce a deterministic generation method that formulates IR as a visual question answering task and enforces deterministic outputs, enabling training-free zero-shot IR. To further enhance performance and efficiency by fine-tuning the model, we design a spatial-aware pooling module that integrates appearance and pairwise spatial cues, and a one-pass deterministic matching method that predicts all candidate interactions in a single forward pass. Extensive experiments on HICO-DET and V-COCO demonstrate that our method achieves superior zero-shot performance, strong cross-dataset generalization, and the flexibility to integrate with any object detectors without retraining. The codes are publicly available at https://github.com/SY-Xuan/DA-HOI.

Hao Tang, Jun Liu, Shuanglin Yan et al.

Due to the scarcity of manually annotated data required for fine-grained video understanding, few-shot fine-grained (FS-FG) action recognition has gained significant attention, with the aim of classifying novel fine-grained action categories with only a few labeled instances. Despite the progress made in FS coarse-grained action recognition, current approaches encounter two challenges when dealing with the fine-grained action categories: the inability to capture subtle action details and the insufficiency of learning from limited data that exhibit high intra-class variance and inter-class similarity. To address these limitations, we propose M$^3$Net, a matching-based framework for FS-FG action recognition, which incorporates \textit{multi-view encoding}, \textit{multi-view matching}, and \textit{multi-view fusion} to facilitate embedding encoding, similarity matching, and decision making across multiple viewpoints. \textit{Multi-view encoding} captures rich contextual details from the intra-frame, intra-video, and intra-episode perspectives, generating customized higher-order embeddings for fine-grained data. \textit{Multi-view matching} integrates various matching functions enabling flexible relation modeling within limited samples to handle multi-scale spatio-temporal variations by leveraging the instance-specific, category-specific, and task-specific perspectives. \textit{Multi-view fusion} consists of matching-predictions fusion and matching-losses fusion over the above views, where the former promotes mutual complementarity and the latter enhances embedding generalizability by employing multi-task collaborative learning. Explainable visualizations and experimental results on three challenging benchmarks demonstrate the superiority of M$^3$Net in capturing fine-grained action details and achieving state-of-the-art performance for FS-FG action recognition.

Peng Xing, Zechao Li

Reconstruction method based on the memory module for visual anomaly detection attempts to narrow the reconstruction error for normal samples while enlarging it for anomalous samples. Unfortunately, the existing memory module is not fully applicable to the anomaly detection task, and the reconstruction error of the anomaly samples remains small. Towards this end, this work proposes a new unsupervised visual anomaly detection method to jointly learn effective normal features and eliminate unfavorable reconstruction errors. Specifically, a novel Partition Memory Bank (PMB) module is proposed to effectively learn and store detailed features with semantic integrity of normal samples. It develops a new partition mechanism and a unique query generation method to preserve the context information and then improves the learning ability of the memory module. The proposed PMB and the skip connection are alternatively explored to make the reconstruction of abnormal samples worse. To obtain more precise anomaly localization results and solve the problem of cumulative reconstruction error, a novel Histogram Error Estimation module is proposed to adaptively eliminate the unfavorable errors by the histogram of the difference image. It improves the anomaly localization performance without increasing the cost. To evaluate the effectiveness of the proposed method for anomaly detection and localization, extensive experiments are conducted on three widely-used anomaly detection datasets. The encouraging performance of the proposed method compared to the recent approaches based on the memory module demonstrates its superiority.

Peng Xing, Hao Tang, Jinhui Tang et al.

Knowledge Distillation-based Anomaly Detection (KDAD) methods rely on the teacher-student paradigm to detect and segment anomalous regions by contrasting the unique features extracted by both networks. However, existing KDAD methods suffer from two main limitations: 1) the student network can effortlessly replicate the teacher network's representations, and 2) the features of the teacher network serve solely as a ``reference standard" and are not fully leveraged. Toward this end, we depart from the established paradigm and instead propose an innovative approach called Asymmetric Distillation Post-Segmentation (ADPS). Our ADPS employs an asymmetric distillation paradigm that takes distinct forms of the same image as the input of the teacher-student networks, driving the student network to learn discriminating representations for anomalous regions. Meanwhile, a customized Weight Mask Block (WMB) is proposed to generate a coarse anomaly localization mask that transfers the distilled knowledge acquired from the asymmetric paradigm to the teacher network. Equipped with WMB, the proposed Post-Segmentation Module (PSM) is able to effectively detect and segment abnormal regions with fine structures and clear boundaries. Experimental results demonstrate that the proposed ADPS outperforms the state-of-the-art methods in detecting and segmenting anomalies. Surprisingly, ADPS significantly improves Average Precision (AP) metric by 9% and 20% on the MVTec AD and KolektorSDD2 datasets, respectively.

Peng Xing, Yanpeng Sun, Zechao Li

Unsupervised anomaly detection is a challenging task in industrial applications since it is impracticable to collect sufficient anomalous samples. In this paper, a novel Self-Supervised Guided Segmentation Framework (SGSF) is proposed by jointly exploring effective generation method of forged anomalous samples and the normal sample features as the guidance information of segmentation for anomaly detection. Specifically, to ensure that the generated forged anomaly samples are conducive to model training, the Saliency Augmentation Module (SAM) is proposed. SAM introduces a saliency map to generate saliency Perlin noise map, and develops an adaptive segmentation strategy to generate irregular masks in the saliency region. Then, the masks are utilized to generate forged anomalous samples as negative samples for training. Unfortunately, the distribution gap between forged and real anomaly samples makes it difficult for models trained based on forged samples to effectively locate real anomalies. Towards this end, the Self-supervised Guidance Network (SGN) is proposed. It leverages the self-supervised module to extract features that are noise-free and contain normal semantic information as the prior knowledge of the segmentation module. The segmentation module with the knowledge of normal patterns segments out the abnormal regions that are different from the guidance features. To evaluate the effectiveness of SGSF for anomaly detection, extensive experiments are conducted on three anomaly detection datasets. The experimental results show that SGSF achieves state-of-the-art anomaly detection results.

Wei Tang, Xuejing Liu, Yanpeng Sun et al.

The Segment Anything Model (SAM) excels at general image segmentation but has limited ability to understand natural language, which restricts its direct application in Referring Expression Segmentation (RES). Toward this end, we propose SSP-SAM, a framework that fully utilizes SAM's segmentation capabilities by integrating a Semantic-Spatial Prompt (SSP) encoder. Specifically, we incorporate both visual and linguistic attention adapters into the SSP encoder, which highlight salient objects within the visual features and discriminative phrases within the linguistic features. This design enhances the referent representation for the prompt generator, resulting in high-quality SSPs that enable SAM to generate precise masks guided by language. Although not specifically designed for Generalized RES (GRES), where the referent may correspond to zero, one, or multiple objects, SSP-SAM naturally supports this more flexible setting without additional modifications. Extensive experiments on widely used RES and GRES benchmarks confirm the superiority of our method. Notably, our approach generates segmentation masks of high quality, achieving strong precision even at strict thresholds such as Pr@0.9. Further evaluation on the PhraseCut dataset demonstrates improved performance in open-vocabulary scenarios compared to existing state-of-the-art RES methods. The code and checkpoints are available at: https://github.com/WayneTomas/SSP-SAM.

Weihao Bo, Yanpeng Sun, Yu Wang et al.

In this paper, we introduce FedMGP, a new paradigm for personalized federated prompt learning in vision-language models. FedMGP equips each client with multiple groups of paired textual and visual prompts, enabling the model to capture diverse, fine-grained semantic and instance-level cues. A diversity loss is introduced to drive each prompt group to specialize in distinct and complementary semantic aspects, ensuring that the groups collectively cover a broader range of local characteristics. During communication, FedMGP employs a dynamic prompt aggregation strategy based on similarity-guided probabilistic sampling: each client computes the cosine similarity between its prompt groups and the global prompts from the previous round, then samples s groups via a softmax-weighted distribution. This soft selection mechanism preferentially aggregates semantically aligned knowledge while still enabling exploration of underrepresented patterns effectively balancing the preservation of common knowledge with client-specific features. Notably, FedMGP maintains parameter efficiency by redistributing a fixed prompt capacity across multiple groups, achieving state-of-the-art performance with the lowest communication parameters among all federated prompt learning methods. Theoretical analysis shows that our dynamic aggregation strategy promotes robust global representation learning by reinforcing shared semantics while suppressing client-specific noise. Extensive experiments demonstrate that FedMGP consistently outperforms prior approaches in both personalization and domain generalization across diverse federated vision-language benchmarks. The code will be released on https://github.com/weihao-bo/FedMGP.git.

Peng Xing, Haofan Wang, Yanpeng Sun et al.

The diffusion model has shown exceptional capabilities in controlled image generation, which has further fueled interest in image style transfer. Existing works mainly focus on training free-based methods (e.g., image inversion) due to the scarcity of specific data. In this study, we present a data construction pipeline for content-style-stylized image triplets that generates and automatically cleanses stylized data triplets. Based on this pipeline, we construct a dataset IMAGStyle, the first large-scale style transfer dataset containing 210k image triplets, available for the community to explore and research. Equipped with IMAGStyle, we propose CSGO, a style transfer model based on end-to-end training, which explicitly decouples content and style features employing independent feature injection. The unified CSGO implements image-driven style transfer, text-driven stylized synthesis, and text editing-driven stylized synthesis. Extensive experiments demonstrate the effectiveness of our approach in enhancing style control capabilities in image generation. Additional visualization and access to the source code can be located on the project page: \url{https://csgo-gen.github.io/}.

Xin Jiang, Jingwen Chen, Yehao Li et al.

Recent advances in subject-driven image generation using diffusion models have attracted considerable attention for their remarkable capabilities in producing high-quality images. Nevertheless, the potential of Visual Autoregressive (VAR) models, despite their unified architecture and efficient inference, remains underexplored. In this work, we present DreamVAR, a novel framework for subject-driven image synthesis built upon a VAR model that employs next-scale prediction. Technically, multi-scale features of the reference subject are first extracted by a visual tokenizer. Instead of interleaving these conditional features with target image tokens across scales, our DreamVAR pre-fills the full subject feature sequence prior to predicting target image tokens. This design simplifies autoregressive dependencies and mitigates the train-test discrepancy in multi-scale conditioning scenario within the VAR paradigm. DreamVAR further incorporates reinforcement learning to jointly enhance semantic alignment and subject consistency. Extensive experiments demonstrate that DreamVAR achieves superior appearance preservation compared to leading diffusion-based methods.

Xin Jiang, Hao Tang, Junyao Gao et al.

Fine-grained visual classification (FGVC) involves categorizing fine subdivisions within a broader category, which poses challenges due to subtle inter-class discrepancies and large intra-class variations. However, prevailing approaches primarily focus on uni-modal visual concepts. Recent advancements in pre-trained vision-language models have demonstrated remarkable performance in various high-level vision tasks, yet the applicability of such models to FGVC tasks remains uncertain. In this paper, we aim to fully exploit the capabilities of cross-modal description to tackle FGVC tasks and propose a novel multimodal prompting solution, denoted as MP-FGVC, based on the contrastive language-image pertaining (CLIP) model. Our MP-FGVC comprises a multimodal prompts scheme and a multimodal adaptation scheme. The former includes Subcategory-specific Vision Prompt (SsVP) and Discrepancy-aware Text Prompt (DaTP), which explicitly highlights the subcategory-specific discrepancies from the perspectives of both vision and language. The latter aligns the vision and text prompting elements in a common semantic space, facilitating cross-modal collaborative reasoning through a Vision-Language Fusion Module (VLFM) for further improvement on FGVC. Moreover, we tailor a two-stage optimization strategy for MP-FGVC to fully leverage the pre-trained CLIP model and expedite efficient adaptation for FGVC. Extensive experiments conducted on four FGVC datasets demonstrate the effectiveness of our MP-FGVC.

Yanpeng Sun, Jiahui Chen, Shan Zhang et al.

In this paper, we propose a novel Visual Reference Prompt (VRP) encoder that empowers the Segment Anything Model (SAM) to utilize annotated reference images as prompts for segmentation, creating the VRP-SAM model. In essence, VRP-SAM can utilize annotated reference images to comprehend specific objects and perform segmentation of specific objects in target image. It is note that the VRP encoder can support a variety of annotation formats for reference images, including \textbf{point}, \textbf{box}, \textbf{scribble}, and \textbf{mask}. VRP-SAM achieves a breakthrough within the SAM framework by extending its versatility and applicability while preserving SAM's inherent strengths, thus enhancing user-friendliness. To enhance the generalization ability of VRP-SAM, the VRP encoder adopts a meta-learning strategy. To validate the effectiveness of VRP-SAM, we conducted extensive empirical studies on the Pascal and COCO datasets. Remarkably, VRP-SAM achieved state-of-the-art performance in visual reference segmentation with minimal learnable parameters. Furthermore, VRP-SAM demonstrates strong generalization capabilities, allowing it to perform segmentation of unseen objects and enabling cross-domain segmentation. The source code and models will be available at https://github.com/syp2ysy/VRP-SAM

Xuejing Liu, Wei Tang, Xinzhe Ni et al.

Text-rich VQA, namely Visual Question Answering based on text recognition in the images, is a cross-modal task that requires both image comprehension and text recognition. In this work, we focus on investigating the advantages and bottlenecks of LLM-based approaches in addressing this problem. To address the above concern, we separate the vision and language modules, where we leverage external OCR models to recognize texts in the image and Large Language Models (LLMs) to answer the question given texts. The whole framework is training-free benefiting from the in-context ability of LLMs. This pipeline achieved superior performance compared to the majority of existing Multimodal Large Language Models (MLLM) on four text-rich VQA datasets. Besides, based on the ablation study, we find that LLM brings stronger comprehension ability and may introduce helpful knowledge for the VQA problem. The bottleneck for LLM to address text-rich VQA problems may primarily lie in visual part. We also combine the OCR module with MLLMs and pleasantly find that the combination of OCR module with MLLM also works. It's worth noting that not all MLLMs can comprehend the OCR information, which provides insights into how to train an MLLM that preserves the abilities of LLM.

Wei Luo, Peng Xing, Yunkang Cao et al.

Unsupervised anomaly detection plays a pivotal role in industrial defect inspection and medical image analysis, with most methods relying on the reconstruction framework. However, these methods may suffer from over-generalization, enabling them to reconstruct anomalies well, which leads to poor detection performance. To address this issue, instead of focusing solely on normality reconstruction, we propose an innovative Uncertainty-Integrated Anomaly Perception and Restoration Attention Network (URA-Net), which explicitly restores abnormal patterns to their corresponding normality. First, unlike traditional image reconstruction methods, we utilize a pre-trained convolutional neural network to extract multi-level semantic features as the reconstruction target. To assist the URA-Net learning to restore anomalies, we introduce a novel feature-level artificial anomaly synthesis module to generate anomalous samples for training. Subsequently, a novel uncertainty-integrated anomaly perception module based on Bayesian neural networks is introduced to learn the distributions of anomalous and normal features. This facilitates the estimation of anomalous regions and ambiguous boundaries, laying the foundation for subsequent anomaly restoration. Then, we propose a novel restoration attention mechanism that leverages global normal semantic information to restore detected anomalous regions, thereby obtaining defect-free restored features. Finally, we employ residual maps between input features and restored features for anomaly detection and localization. The comprehensive experimental results on two industrial datasets, MVTec AD and BTAD, along with a medical image dataset, OCT-2017, unequivocally demonstrate the effectiveness and superiority of the proposed method.

Wei Tang, Liang Li, Xuejing Liu et al.

Visual grounding (VG) aims to locate a specific target in an image based on a given language query. The discriminative information from context is important for distinguishing the target from other objects, particularly for the targets that have the same category as others. However, most previous methods underestimate such information. Moreover, they are usually designed for the standard scene (without any novel object), which limits their generalization to the open-vocabulary scene. In this paper, we propose a novel framework with context disentangling and prototype inheriting for robust visual grounding to handle both scenes. Specifically, the context disentangling disentangles the referent and context features, which achieves better discrimination between them. The prototype inheriting inherits the prototypes discovered from the disentangled visual features by a prototype bank to fully utilize the seen data, especially for the open-vocabulary scene. The fused features, obtained by leveraging Hadamard product on disentangled linguistic and visual features of prototypes to avoid sharp adjusting the importance between the two types of features, are then attached with a special token and feed to a vision Transformer encoder for bounding box regression. Extensive experiments are conducted on both standard and open-vocabulary scenes. The performance comparisons indicate that our method outperforms the state-of-the-art methods in both scenarios. {The code is available at https://github.com/WayneTomas/TransCP.

Ziye Fang, Xin Jiang, Hao Tang et al.

In the field of intelligent multimedia analysis, ultra-fine-grained visual categorization (Ultra-FGVC) plays a vital role in distinguishing intricate subcategories within broader categories. However, this task is inherently challenging due to the complex granularity of category subdivisions and the limited availability of data for each category. To address these challenges, this work proposes CSDNet, a pioneering framework that effectively explores contrastive learning and self-distillation to learn discriminative representations specifically designed for Ultra-FGVC tasks. CSDNet comprises three main modules: Subcategory-Specific Discrepancy Parsing (SSDP), Dynamic Discrepancy Learning (DDL), and Subcategory-Specific Discrepancy Transfer (SSDT), which collectively enhance the generalization of deep models across instance, feature, and logit prediction levels. To increase the diversity of training samples, the SSDP module introduces adaptive augmented samples to spotlight subcategory-specific discrepancies. Simultaneously, the proposed DDL module stores historical intermediate features by a dynamic memory queue, which optimizes the feature learning space through iterative contrastive learning. Furthermore, the SSDT module effectively distills subcategory-specific discrepancies knowledge from the inherent structure of limited training data using a self-distillation paradigm at the logit prediction level. Experimental results demonstrate that CSDNet outperforms current state-of-the-art Ultra-FGVC methods, emphasizing its powerful efficacy and adaptability in addressing Ultra-FGVC tasks.

Henghao Zhao, Kevin Qinghong Lin, Rui Yan et al.

Video moment retrieval and highlight detection have received attention in the current era of video content proliferation, aiming to localize moments and estimate clip relevances based on user-specific queries. Given that the video content is continuous in time, there is often a lack of clear boundaries between temporal events in a video. This boundary ambiguity makes it challenging for the model to learn text-video clip correspondences, resulting in the subpar performance of existing methods in predicting target segments. To alleviate this problem, we propose to solve the two tasks jointly from the perspective of denoising generation. Moreover, the target boundary can be localized clearly by iterative refinement from coarse to fine. Specifically, a novel framework, DiffusionVMR, is proposed to redefine the two tasks as a unified conditional denoising generation process by combining the diffusion model. During training, Gaussian noise is added to corrupt the ground truth, with noisy candidates produced as input. The model is trained to reverse this noise addition process. In the inference phase, DiffusionVMR initiates directly from Gaussian noise and progressively refines the proposals from the noise to the meaningful output. Notably, the proposed DiffusionVMR inherits the advantages of diffusion models that allow for iteratively refined results during inference, enhancing the boundary transition from coarse to fine. Furthermore, the training and inference of DiffusionVMR are decoupled. An arbitrary setting can be used in DiffusionVMR during inference without consistency with the training phase. Extensive experiments conducted on five widely-used benchmarks (i.e., QVHighlight, Charades-STA, TACoS, YouTubeHighlights and TVSum) across two tasks (moment retrieval and/or highlight detection) demonstrate the effectiveness and flexibility of the proposed DiffusionVMR.

Enquan Yang, Peng Xing, Hanyang Sun et al.

Industrial anomaly detection achieves progress thanks to datasets such as MVTec-AD and VisA. However, they suffer from limitations in terms of the number of defect samples, types of defects, and availability of real-world scenes. These constraints inhibit researchers from further exploring the performance of industrial detection with higher accuracy. To this end, we propose a new large-scale anomaly detection dataset called 3CAD, which is derived from real 3C production lines. Specifically, the proposed 3CAD includes eight different types of manufactured parts, totaling 27,039 high-resolution images labeled with pixel-level anomalies. The key features of 3CAD are that it covers anomalous regions of different sizes, multiple anomaly types, and the possibility of multiple anomalous regions and multiple anomaly types per anomaly image. This is the largest and first anomaly detection dataset dedicated to 3C product quality control for community exploration and development. Meanwhile, we introduce a simple yet effective framework for unsupervised anomaly detection: a Coarse-to-Fine detection paradigm with Recovery Guidance (CFRG). To detect small defect anomalies, the proposed CFRG utilizes a coarse-to-fine detection paradigm. Specifically, we utilize a heterogeneous distillation model for coarse localization and then fine localization through a segmentation model. In addition, to better capture normal patterns, we introduce recovery features as guidance. Finally, we report the results of our CFRG framework and popular anomaly detection methods on the 3CAD dataset, demonstrating strong competitiveness and providing a highly challenging benchmark to promote the development of the anomaly detection field. Data and code are available: https://github.com/EnquanYang2022/3CAD.

Yanpeng Sun, Jing Hao, Ke Zhu et al.

Training Large Multimodality Models (LMMs) relies on descriptive image caption that connects image and language. Existing methods either distill the caption from the LMM models or construct the captions from the internet images or by human. We propose to leverage off-the-shelf visual specialists, which were trained from annotated images initially not for image captioning, for enhancing the image caption. Our approach, named DCE, explores object low-level and fine-grained attributes (e.g., depth, emotion and fine-grained categories) and object relations (e.g., relative location and human-object-interaction (HOI)), and combine the attributes into the descriptive caption. Experiments demonstrate that such visual specialists are able to improve the performance for visual understanding tasks as well as reasoning that benefits from more accurate visual understanding. We will release the source code and the pipeline so that other visual specialists are easily combined into the pipeline. The complete source code of DCE pipeline and datasets will be available at \url{https://github.com/syp2ysy/DCE}.

Deng Xu, Chao Zhang, Zechao Li et al.

Tensor-based multi-view clustering has recently received significant attention due to its exceptional ability to explore cross-view high-order correlations. However, most existing methods still encounter some limitations. (1) Most of them explore the correlations among different affinity matrices, making them unscalable to large-scale data. (2) Although some methods address it by introducing bipartite graphs, they may result in sub-optimal solutions caused by an unstable anchor selection process. (3) They generally ignore the negative impact of latent semantic-unrelated information in each view. To tackle these issues, we propose a new approach termed fast Disentangled Slim Tensor Learning (DSTL) for multi-view clustering . Instead of focusing on the multi-view graph structures, DSTL directly explores the high-order correlations among multi-view latent semantic representations based on matrix factorization. To alleviate the negative influence of feature redundancy, inspired by robust PCA, DSTL disentangles the latent low-dimensional representation into a semantic-unrelated part and a semantic-related part for each view. Subsequently, two slim tensors are constructed with tensor-based regularization. To further enhance the quality of feature disentanglement, the semantic-related representations are aligned across views through a consensus alignment indicator. Our proposed model is computationally efficient and can be solved effectively. Extensive experiments demonstrate the superiority and efficiency of DSTL over state-of-the-art approaches. The code of DSTL is available at https://github.com/dengxu-nju/DSTL.

Wei Tang, Yanpeng Sun, Qinying Gu et al.

Although Multimodal Large Language Models (MLLMs) excel at various image-related tasks, they encounter challenges in precisely aligning coordinates with spatial information within images, particularly in position-aware tasks such as visual grounding. This limitation arises from two key factors. First, MLLMs lack explicit spatial references, making it difficult to associate textual descriptions with precise image locations. Second, their feature extraction processes prioritize global context over fine-grained spatial details, leading to weak localization capability. To address these issues, we introduce VPP-LLaVA, an MLLM enhanced with Visual Position Prompt (VPP) to improve its grounding capability. VPP-LLaVA integrates two complementary mechanisms: the global VPP overlays a learnable, axis-like tensor onto the input image to provide structured spatial cues, while the local VPP incorporates position-aware queries to support fine-grained localization.To effectively train our model with spatial guidance, we further introduce VPP-SFT, a curated dataset of 0.6M high-quality visual grounding samples. Designed in a compact format, it enables efficient training and is significantly smaller than datasets used by other MLLMs (e.g., ~21M samples in MiniGPT-v2), yet still provides a strong performance boost. The resulting model, VPP-LLaVA, not only achieves state-of-the-art results on standard visual grounding benchmarks but also demonstrates strong zero-shot generalization to challenging unseen datasets. The code and dataset are available at https://github.com/WayneTomas/VPP-LLaVA.

Yuntian Bo, Tao Zhou, Zechao Li et al.

Cross-domain few-shot medical image segmentation (CD-FSMIS) offers a promising and data-efficient solution for medical applications where annotations are severely scarce and multimodal analysis is required. However, existing methods typically filter out domain-specific information to improve generalization, which inadvertently limits cross-domain performance and degrades source-domain accuracy. To address this, we present Contrastive Graph Modeling (C-Graph), a framework that leverages the structural consistency of medical images as a reliable domain-transferable prior. We represent image features as graphs, with pixels as nodes and semantic affinities as edges. A Structural Prior Graph (SPG) layer is proposed to capture and transfer target-category node dependencies and enable global structure modeling through explicit node interactions. Building upon SPG layers, we introduce a Subgraph Matching Decoding (SMD) mechanism that exploits semantic relations among nodes to guide prediction. Furthermore, we design a Confusion-minimizing Node Contrast (CNC) loss to mitigate node ambiguity and subgraph heterogeneity by contrastively enhancing node discriminability in the graph space. Our method significantly outperforms prior CD-FSMIS approaches across multiple cross-domain benchmarks, achieving state-of-the-art performance while simultaneously preserving strong segmentation accuracy on the source domain.

Wei Tang, Yanpeng Sun, Shan Zhang et al.

Recent reinforcement-learning frameworks for visual perception policy have begun to incorporate intermediate reasoning chains expressed in natural language. Empirical observations indicate that such purely linguistic intermediate reasoning often reduces performance on perception tasks. We argue that the core issue lies not in reasoning per se but in the form of reasoning: while these chains perform semantic reasoning in an unstructured linguistic space, visual perception requires reasoning in a spatial and object-centric space. In response, we introduce Artemis, a perception-policy learning framework that performs structured proposal-based reasoning, where each intermediate step is represented as a (label, bounding-box) pair capturing a verifiable visual state. This design enables explicit tracking of intermediate states, direct supervision for proposal quality, and avoids ambiguity introduced by language-based reasoning. Artemis is built on Qwen2.5-VL-3B, achieves strong performance on grounding and detection task and exhibits substantial generalization to counting and geometric-perception tasks. The consistent improvements across these diverse settings confirm that aligning reasoning with spatial representations enhances perception-policy learning. Owing to its strengthened visual reasoning, Artemis also achieves competitive performance on general MLLM benchmarks, illustrating that spatially grounded reasoning provides a principled route toward scalable and general perception policies.

Jiaqi Ma, Guo-Sen Xie, Fang Zhao et al.

Few-shot learning aims to recognize novel concepts by leveraging prior knowledge learned from a few samples. However, for visually intensive tasks such as few-shot semantic segmentation, pixel-level annotations are time-consuming and costly. Therefore, in this paper, we utilize the more challenging image-level annotations and propose an adaptive frequency-aware network (AFANet) for weakly-supervised few-shot semantic segmentation (WFSS). Specifically, we first propose a cross-granularity frequency-aware module (CFM) that decouples RGB images into high-frequency and low-frequency distributions and further optimizes semantic structural information by realigning them. Unlike most existing WFSS methods using the textual information from the multi-modal language-vision model, e.g., CLIP, in an offline learning manner, we further propose a CLIP-guided spatial-adapter module (CSM), which performs spatial domain adaptive transformation on textual information through online learning, thus providing enriched cross-modal semantic information for CFM. Extensive experiments on the Pascal-5\textsuperscript{i} and COCO-20\textsuperscript{i} datasets demonstrate that AFANet has achieved state-of-the-art performance. The code is available at https://github.com/jarch-ma/AFANet.

Weihao Bo, Shan Zhang, Yanpeng Sun et al.

MLLMs exhibit strong reasoning on isolated queries, yet they operate de novo -- solving each problem independently and often repeating the same mistakes. Existing memory-augmented agents mainly store past trajectories for reuse. However, trajectory-based memory suffers from brevity bias, gradually losing essential domain knowledge. More critically, even in truly multimodal problem-solving settings, it records only a single-modality trace of past behavior, failing to preserve how visual attention and logical reasoning jointly contributed to the solution. This is fundamentally misaligned with human cognition: semantic memory is both multimodal and integrated, preserving visual and abstract knowledge through coordinated but distinct representational streams. We thus introduce ViLoMem, a dual-stream memory framework that constructs compact, schema-based memory. It separately encodes visual distraction patterns and logical reasoning errors, enabling MLLMs to learn from their successful and failed experiences. Following a grow-and-refine principle, the system incrementally accumulates and updates multimodal semantic knowledge -- preserving stable, generalizable strategies while avoiding catastrophic forgetting. Across six multimodal benchmarks, ViLoMem consistently improves pass@1 accuracy and substantially reduces repeated visual and logical errors. Ablations confirm the necessity of dual-stream memory with explicit distraction--hallucination separation, demonstrating the value of error-aware multimodal memory for lifelong and cross-domain agentic learning. Our project page will be available at https://weihao-bo.github.io/ViLoMeo-page.

Chengwei Wu, Jiapu Wang, Mingyang Gao et al.

Large Language Models (LLMs) have achieved remarkable success across a wide range of natural language processing tasks. However, Chinese LLMs face unique challenges, primarily due to the dominance of unstructured free text and the lack of structured representations in Chinese corpora. While existing benchmarks for LLMs partially assess Chinese LLMs, they are still predominantly English-centric and fail to address the unique linguistic characteristics of Chinese, lacking structured datasets essential for robust evaluation. To address these challenges, we present a Comprehensive Benchmark for Evaluating Chinese Large Language Models (CB-ECLLM) based on the newly constructed Chinese Data-Text Pair (CDTP) dataset. Specifically, CDTP comprises over 7 million aligned text pairs, each consisting of unstructured text coupled with one or more corresponding triples, alongside a total of 15 million triples spanning four critical domains. The core contributions of CDTP are threefold: (i) enriching Chinese corpora with high-quality structured information; (ii) enabling fine-grained evaluation tailored to knowledge-driven tasks; and (iii) supporting multi-task fine-tuning to assess generalization and robustness across scenarios, including Knowledge Graph Completion, Triple-to-Text generation, and Question Answering. Furthermore, we conduct rigorous evaluations through extensive experiments and ablation studies to assess the effectiveness, Supervised Fine-Tuning (SFT), and robustness of the benchmark. To support reproducible research, we offer an open-source codebase and outline potential directions for future investigations based on our insights.

Jiaqi Ma, Guo-Sen Xie, Fang Zhao et al.

Meta-learning aims to uniformly sample homogeneous support-query pairs, characterized by the same categories and similar attributes, and extract useful inductive biases through identical network architectures. However, this identical network design results in over-semantic homogenization. To address this, we propose a novel homologous but heterogeneous network. By treating support-query pairs as dual perspectives, we introduce heterogeneous visual aggregation (HA) modules to enhance complementarity while preserving semantic commonality. To further reduce semantic noise and amplify the uniqueness of heterogeneous semantics, we design a heterogeneous transfer (HT) module. Finally, we propose heterogeneous CLIP (HC) textual information to enhance the generalization capability of multimodal models. In the weakly-supervised few-shot semantic segmentation (WFSS) task, with only 1/24 of the parameters of existing state-of-the-art models, TLG achieves a 13.2\% improvement on Pascal-5\textsuperscript{i} and a 9.7\% improvement on COCO-20\textsuperscript{i}. To the best of our knowledge, TLG is also the first weakly supervised (image-level) model that outperforms fully supervised (pixel-level) models under the same backbone architectures. The code is available at https://github.com/jarch-ma/TLG.

Xin Jiang, Hao Tang, Rui Yan et al.

Fine-grained image retrieval (FGIR) is to learn visual representations that distinguish visually similar objects while maintaining generalization. Existing methods propose to generate discriminative features, but rarely consider the particularity of the FGIR task itself. This paper presents a meticulous analysis leading to the proposal of practical guidelines to identify subcategory-specific discrepancies and generate discriminative features to design effective FGIR models. These guidelines include emphasizing the object (G1), highlighting subcategory-specific discrepancies (G2), and employing effective training strategy (G3). Following G1 and G2, we design a novel Dual Visual Filtering mechanism for the plain visual transformer, denoted as DVF, to capture subcategory-specific discrepancies. Specifically, the dual visual filtering mechanism comprises an object-oriented module and a semantic-oriented module. These components serve to magnify objects and identify discriminative regions, respectively. Following G3, we implement a discriminative model training strategy to improve the discriminability and generalization ability of DVF. Extensive analysis and ablation studies confirm the efficacy of our proposed guidelines. Without bells and whistles, the proposed DVF achieves state-of-the-art performance on three widely-used fine-grained datasets in closed-set and open-set settings.

Yanpeng Sun, Huaxin Zhang, Qiang Chen et al.

We focus on improving the visual understanding capability for boosting the vision-language models. We propose \textbf{Arcana}, a multiModal language model, which introduces two crucial techniques. First, we present Multimodal LoRA (MM-LoRA), a module designed to enhance the decoder. Unlike traditional language-driven decoders, MM-LoRA consists of two parallel LoRAs -- one for vision and one for language -- each with its own parameters. This disentangled parameters design allows for more specialized learning in each modality and better integration of multimodal information. Second, we introduce the Query Ladder adapter (QLadder) to improve the visual encoder. QLadder employs a learnable ``\textit{ladder}'' structure to deeply aggregates the intermediate representations from the frozen pretrained visual encoder (e.g., CLIP image encoder). This enables the model to learn new and informative visual features, as well as remaining the powerful capabilities of the pretrained visual encoder. These techniques collectively enhance Arcana's visual perception power, enabling it to leverage improved visual information for more accurate and contextually relevant outputs across various multimodal scenarios. Extensive experiments and ablation studies demonstrate the effectiveness and generalization capability of our Arcana. The code and re-annotated data are available at \url{https://arcana-project-page.github.io}.

Henghao Zhao, Ge-Peng Ji, Rui Yan et al.

The core challenge in video understanding lies in perceiving dynamic content changes over time. However, multimodal large language models struggle with temporal-sensitive video tasks, which requires generating timestamps to mark the occurrence of specific events. Existing strategies require MLLMs to generate absolute or relative timestamps directly. We have observed that those MLLMs tend to rely more on language patterns than visual cues when generating timestamps, affecting their performance. To address this problem, we propose VideoExpert, a general-purpose MLLM suitable for several temporal-sensitive video tasks. Inspired by the expert concept, VideoExpert integrates two parallel modules: the Temporal Expert and the Spatial Expert. The Temporal Expert is responsible for modeling time sequences and performing temporal grounding. It processes high-frame-rate yet compressed tokens to capture dynamic variations in videos and includes a lightweight prediction head for precise event localization. The Spatial Expert focuses on content detail analysis and instruction following. It handles specially designed spatial tokens and language input, aiming to generate content-related responses. These two experts collaborate seamlessly via a special token, ensuring coordinated temporal grounding and content generation. Notably, the Temporal and Spatial Experts maintain independent parameter sets. By offloading temporal grounding from content generation, VideoExpert prevents text pattern biases in timestamp predictions. Moreover, we introduce a Spatial Compress module to obtain spatial tokens. This module filters and compresses patch tokens while preserving key information, delivering compact yet detail-rich input for the Spatial Expert. Extensive experiments demonstrate the effectiveness and versatility of the VideoExpert.

Jiawei Gu, Ziyue Qiao, Zechao Li

The task of graph-level out-of-distribution (OOD) detection is crucial for deploying graph neural networks in real-world settings. In this paper, we observe a significant difference in the relationship between the largest and second-largest eigenvalues of the Laplacian matrix for in-distribution (ID) and OOD graph samples: \textit{OOD samples often exhibit anomalous spectral gaps (the difference between the largest and second-largest eigenvalues)}. This observation motivates us to propose SpecGap, an effective post-hoc approach for OOD detection on graphs. SpecGap adjusts features by subtracting the component associated with the second-largest eigenvalue, scaled by the spectral gap, from the high-level features (i.e., $\mathbf{X}-\left(λ_n-λ_{n-1}\right) \mathbf{u}_{n-1} \mathbf{v}_{n-1}^T$). SpecGap achieves state-of-the-art performance across multiple benchmark datasets. We present extensive ablation studies and comprehensive theoretical analyses to support our empirical results. As a parameter-free post-hoc method, SpecGap can be easily integrated into existing graph neural network models without requiring any additional training or model modification.

Meiqi Cao, Xiangbo Shu, Jiachao Zhang et al.

Event-based Action Recognition (EAR) possesses the advantages of high-temporal resolution capturing and privacy preservation compared with traditional action recognition. Current leading EAR solutions typically follow two regimes: project unconstructed event streams into dense constructed event frames and adopt powerful frame-specific networks, or employ lightweight point-specific networks to handle sparse unconstructed event points directly. However, such two regimes are blind to a fundamental issue: failing to accommodate the unique dense temporal and sparse spatial properties of asynchronous event data. In this article, we present a synergy-aware framework, i.e., EventCrab, that adeptly integrates the "lighter" frame-specific networks for dense event frames with the "heavier" point-specific networks for sparse event points, balancing accuracy and efficiency. Furthermore, we establish a joint frame-text-point representation space to bridge distinct event frames and points. In specific, to better exploit the unique spatiotemporal relationships inherent in asynchronous event points, we devise two strategies for the "heavier" point-specific embedding: i) a Spiking-like Context Learner (SCL) that extracts contextualized event points from raw event streams. ii) an Event Point Encoder (EPE) that further explores event-point long spatiotemporal features in a Hilbert-scan way. Experiments on four datasets demonstrate the significant performance of our proposed EventCrab, particularly gaining improvements of 5.17% on SeAct and 7.01% on HARDVS.

Yu Liu, Hao Tang, Haiqi Zhang et al.

Out-of-distribution (OOD) detection is crucial for ensuring the reliability and safety of machine learning models in real-world applications. While zero-shot OOD detection, which requires no training on in-distribution (ID) data, has become feasible with the emergence of vision-language models like CLIP, existing methods primarily focus on semantic matching and fail to fully capture distributional discrepancies. To address these limitations, we propose OT-DETECTOR, a novel framework that employs Optimal Transport (OT) to quantify both semantic and distributional discrepancies between test samples and ID labels. Specifically, we introduce cross-modal transport mass and transport cost as semantic-wise and distribution-wise OOD scores, respectively, enabling more robust detection of OOD samples. Additionally, we present a semantic-aware content refinement (SaCR) module, which utilizes semantic cues from ID labels to amplify the distributional discrepancy between ID and hard OOD samples. Extensive experiments on several benchmarks demonstrate that OT-DETECTOR achieves state-of-the-art performance across various OOD detection tasks, particularly in challenging hard-OOD scenarios.

Jiawei Gu, Ziyue Qiao, Zechao Li

Out-of-Distribution (OOD) detection is critical for safely deploying deep models in open-world environments, where inputs may lie outside the training distribution. During inference on a model trained exclusively with In-Distribution (ID) data, we observe a salient gradient phenomenon: around an ID sample, the local gradient directions for "enhancing" that sample's predicted class remain relatively consistent, whereas OOD samples--unseen in training--exhibit disorganized or conflicting gradient directions in the same neighborhood. Motivated by this observation, we propose an inference-stage technique to short-circuit those feature coordinates that spurious gradients exploit to inflate OOD confidence, while leaving ID classification largely intact. To circumvent the expense of recomputing the logits after this gradient short-circuit, we further introduce a local first-order approximation that accurately captures the post-modification outputs without a second forward pass. Experiments on standard OOD benchmarks show our approach yields substantial improvements. Moreover, the method is lightweight and requires minimal changes to the standard inference pipeline, offering a practical path toward robust OOD detection in real-world applications.

Shiyu Xuan, Zechao Li, Jinhui Tang

Multi-modal object tracking integrates auxiliary modalities such as depth, thermal infrared, event flow, and language to provide additional information beyond RGB images, showing great potential in improving tracking stabilization in complex scenarios. Existing methods typically start from an RGB-based tracker and learn to understand auxiliary modalities only from training data. Constrained by the limited multi-modal training data, the performance of these methods is unsatisfactory. To alleviate this limitation, this work proposes a unified multi-modal tracker Diff-MM by exploiting the multi-modal understanding capability of the pre-trained text-to-image generation model. Diff-MM leverages the UNet of pre-trained Stable Diffusion as a tracking feature extractor through the proposed parallel feature extraction pipeline, which enables pairwise image inputs for object tracking. We further introduce a multi-modal sub-module tuning method that learns to gain complementary information between different modalities. By harnessing the extensive prior knowledge in the generation model, we achieve a unified tracker with uniform parameters for RGB-N/D/T/E tracking. Experimental results demonstrate the promising performance of our method compared with recently proposed trackers, e.g., its AUC outperforms OneTracker by 8.3% on TNL2K.

Ziqiang Li, Jun Li, Lizhi Xiong et al.

Text-to-image diffusion models have made significant advancements in generating high-quality, diverse images from text prompts. However, the inherent limitations of textual signals often prevent these models from fully capturing specific concepts, thereby reducing their controllability. To address this issue, several approaches have incorporated personalization techniques, utilizing reference images to mine visual concept representations that complement textual inputs and enhance the controllability of text-to-image diffusion models. Despite these advances, a comprehensive, systematic exploration of visual concept mining remains limited. In this paper, we categorize existing research into four key areas: Concept Learning, Concept Erasing, Concept Decomposition, and Concept Combination. This classification provides valuable insights into the foundational principles of Visual Concept Mining (VCM) techniques. Additionally, we identify key challenges and propose future research directions to propel this important and interesting field forward.

Zeren Sun, Yazhou Yao, Tongliang Liu et al.

Label noise is pervasive in various real-world scenarios, posing challenges in supervised deep learning. Deep networks are vulnerable to such label-corrupted samples due to the memorization effect. One major stream of previous methods concentrates on identifying clean data for training. However, these methods often neglect imbalances in label noise across different mini-batches and devote insufficient attention to out-of-distribution noisy data. To this end, we propose a noise-robust method named Jo-SNC (\textbf{Jo}int sample selection and model regularization based on \textbf{S}elf- and \textbf{N}eighbor-\textbf{C}onsistency). Specifically, we propose to employ the Jensen-Shannon divergence to measure the ``likelihood'' of a sample being clean or out-of-distribution. This process factors in the nearest neighbors of each sample to reinforce the reliability of clean sample identification. We design a self-adaptive, data-driven thresholding scheme to adjust per-class selection thresholds. While clean samples undergo conventional training, detected in-distribution and out-of-distribution noisy samples are trained following partial label learning and negative learning, respectively. Finally, we advance the model performance further by proposing a triplet consistency regularization that promotes self-prediction consistency, neighbor-prediction consistency, and feature consistency. Extensive experiments on various benchmark datasets and comprehensive ablation studies demonstrate the effectiveness and superiority of our approach over existing state-of-the-art methods.

Ling Xing, Alex Jinpeng Wang, Rui Yan et al.

People see text. Humans read by recognizing words as visual objects, including their shapes, layouts, and patterns, before connecting them to meaning, which enables us to handle typos, distorted fonts, and various scripts effectively. Modern large language models (LLMs), however, rely on subword tokenization, fragmenting text into pieces from a fixed vocabulary. While effective for high-resource languages, this approach over-segments low-resource languages, yielding long, linguistically meaningless sequences and inflating computation. In this work, we challenge this entrenched paradigm and move toward a vision-centric alternative. Our method, SeeTok, renders text as images (visual-text) and leverages pretrained multimodal LLMs to interpret them, reusing strong OCR and text-vision alignment abilities learned from large-scale multimodal training. Across three different language tasks, SeeTok matches or surpasses subword tokenizers while requiring 4.43 times fewer tokens and reducing FLOPs by 70.5%, with additional gains in cross-lingual generalization, robustness to typographic noise, and linguistic hierarchy. SeeTok signals a shift from symbolic tokenization to human-like visual reading, and takes a step toward more natural and cognitively inspired language models.

Xin Jiang, Meiqi Cao, Hao Tang et al.

Fine-Grained Image Retrieval~(FGIR) faces challenges in learning discriminative visual representations to retrieve images with similar fine-grained features. Current leading FGIR solutions typically follow two regimes: enforce pairwise similarity constraints in the semantic embedding space, or incorporate a localization sub-network to fine-tune the entire model. However, such two regimes tend to overfit the training data while forgetting the knowledge gained from large-scale pre-training, thus reducing their generalization ability. In this paper, we propose a Dual-Vision Adaptation (DVA) approach for FGIR, which guides the frozen pre-trained model to perform FGIR through collaborative sample and feature adaptation. Specifically, we design Object-Perceptual Adaptation, which modifies input samples to help the pre-trained model perceive critical objects and elements within objects that are helpful for category prediction. Meanwhile, we propose In-Context Adaptation, which introduces a small set of parameters for feature adaptation without modifying the pre-trained parameters. This makes the FGIR task using these adjusted features closer to the task solved during the pre-training. Additionally, to balance retrieval efficiency and performance, we propose Discrimination Perception Transfer to transfer the discriminative knowledge in the object-perceptual adaptation to the image encoder using the knowledge distillation mechanism. Extensive experiments show that DVA has fewer learnable parameters and performs well on three in-distribution and three out-of-distribution fine-grained datasets.

Hao Tang, Zechao Li, Dong Zhang et al.

RGB-Thermal Salient Object Detection aims to pinpoint prominent objects within aligned pairs of visible and thermal infrared images. Traditional encoder-decoder architectures, while designed for cross-modality feature interactions, may not have adequately considered the robustness against noise originating from defective modalities. Inspired by hierarchical human visual systems, we propose the ConTriNet, a robust Confluent Triple-Flow Network employing a Divide-and-Conquer strategy. Specifically, ConTriNet comprises three flows: two modality-specific flows explore cues from RGB and Thermal modalities, and a third modality-complementary flow integrates cues from both modalities. ConTriNet presents several notable advantages. It incorporates a Modality-induced Feature Modulator in the modality-shared union encoder to minimize inter-modality discrepancies and mitigate the impact of defective samples. Additionally, a foundational Residual Atrous Spatial Pyramid Module in the separated flows enlarges the receptive field, allowing for the capture of multi-scale contextual information. Furthermore, a Modality-aware Dynamic Aggregation Module in the modality-complementary flow dynamically aggregates saliency-related cues from both modality-specific flows. Leveraging the proposed parallel triple-flow framework, we further refine saliency maps derived from different flows through a flow-cooperative fusion strategy, yielding a high-quality, full-resolution saliency map for the final prediction. To evaluate the robustness and stability of our approach, we collect a comprehensive RGB-T SOD benchmark, VT-IMAG, covering various real-world challenging scenarios. Extensive experiments on public benchmarks and our VT-IMAG dataset demonstrate that ConTriNet consistently outperforms state-of-the-art competitors in both common and challenging scenarios.

Xin Jiang, Xu Cheng, Zechao Li

Pre-training has exhibited notable benefits to downstream tasks by boosting accuracy and speeding up convergence, but the exact reasons for these benefits still remain unclear. To this end, we propose to quantitatively and explicitly explain effects of pre-training on the downstream task from a novel game-theoretic view, which also sheds new light into the learning behavior of deep neural networks (DNNs). Specifically, we extract and quantify the knowledge encoded by the pre-trained model, and further track the changes of such knowledge during the fine-tuning process. Interestingly, we discover that only a small amount of pre-trained model's knowledge is preserved for the inference of downstream tasks. However, such preserved knowledge is very challenging for a model training from scratch to learn. Thus, with the help of this exclusively learned and useful knowledge, the model fine-tuned from pre-training usually achieves better performance than the model training from scratch. Besides, we discover that pre-training can guide the fine-tuned model to learn target knowledge for the downstream task more directly and quickly, which accounts for the faster convergence of the fine-tuned model.

Peng Xing, Dong Zhang, Jinhui Tang et al.

Anomaly detection (AD) has been extensively studied and applied in a wide range of scenarios in the recent past. However, there are still gaps between achieved and desirable levels of recognition accuracy for making AD for practical applications. In this paper, we start from an insightful analysis of two types of fundamental yet representative failure cases in the baseline model, and reveal reasons that hinder current AD methods from achieving a higher recognition accuracy. Specifically, by Case-1, we found that the main reasons detrimental to current AD methods is that the inputs to the recovery model contain a large number of detailed features to be recovered, which leads to the normal/abnormal area has-not/has been recovered into its original state. By Case-2, we surprisingly found that the abnormal area that cannot be recognized in image-level representations can be easily recognized in the feature-level representation. Based on the above observations, we propose a novel Recover-then-Discriminate (ReDi) framework for AD. ReDi takes a self-generated feature map and a selected prompted image as explicit input information to solve problems in case-1. Concurrently, a feature-level discriminative network is proposed to enhance abnormal differences between the recovered representation and the input representation. Extensive experimental results on two popular yet challenging AD datasets validate that ReDi achieves the new state-of-the-art accuracy.

Peng Xing, Ning Wang, Jianbo Ouyang et al.

The remarkable advancement in text-to-image generation models significantly boosts the research in ID customization generation. However, existing personalization methods cannot simultaneously satisfy high fidelity and high-efficiency requirements. Their main bottleneck lies in the prompt image encoder, which produces weak alignment signals with the text-to-image model and significantly increased model size. Towards this end, we propose a lightweight Inv-Adapter, which first extracts diffusion-domain representations of ID images utilizing a pre-trained text-to-image model via DDIM image inversion, without additional image encoder. Benefiting from the high alignment of the extracted ID prompt features and the intermediate features of the text-to-image model, we then embed them efficiently into the base text-to-image model by carefully designing a lightweight attention adapter. We conduct extensive experiments to assess ID fidelity, generation loyalty, speed, and training parameters, all of which show that the proposed Inv-Adapter is highly competitive in ID customization generation and model scale.

Tao Chen, Yazhou Yao, Xingguo Huang et al.

The image-level label has prevailed in weakly supervised semantic segmentation tasks due to its easy availability. Since image-level labels can only indicate the existence or absence of specific categories of objects, visualization-based techniques have been widely adopted to provide object location clues. Considering class activation maps (CAMs) can only locate the most discriminative part of objects, recent approaches usually adopt an expansion strategy to enlarge the activation area for more integral object localization. However, without proper constraints, the expanded activation will easily intrude into the background region. In this paper, we propose spatial structure constraints (SSC) for weakly supervised semantic segmentation to alleviate the unwanted object over-activation of attention expansion. Specifically, we propose a CAM-driven reconstruction module to directly reconstruct the input image from deep CAM features, which constrains the diffusion of last-layer object attention by preserving the coarse spatial structure of the image content. Moreover, we propose an activation self-modulation module to refine CAMs with finer spatial structure details by enhancing regional consistency. Without external saliency models to provide background clues, our approach achieves 72.7\% and 47.0\% mIoU on the PASCAL VOC 2012 and COCO datasets, respectively, demonstrating the superiority of our proposed approach.

Jian Zhao, Jianan Li, Lei Jin et al.

The 3rd Anti-UAV Workshop & Challenge aims to encourage research in developing novel and accurate methods for multi-scale object tracking. The Anti-UAV dataset used for the Anti-UAV Challenge has been publicly released. There are two main differences between this year's competition and the previous two. First, we have expanded the existing dataset, and for the first time, released a training set so that participants can focus on improving their models. Second, we set up two tracks for the first time, i.e., Anti-UAV Tracking and Anti-UAV Detection & Tracking. Around 76 participating teams from the globe competed in the 3rd Anti-UAV Challenge. In this paper, we provide a brief summary of the 3rd Anti-UAV Workshop & Challenge including brief introductions to the top three methods in each track. The submission leaderboard will be reopened for researchers that are interested in the Anti-UAV challenge. The benchmark dataset and other information can be found at: https://anti-uav.github.io/.

Yanpeng Sun, Zechao Li

The pixel-wise dense prediction tasks based on weakly supervisions currently use Class Attention Maps (CAM) to generate pseudo masks as ground-truth. However, the existing methods typically depend on the painstaking training modules, which may bring in grinding computational overhead and complex training procedures. In this work, the semantic structure aware inference (SSA) is proposed to explore the semantic structure information hidden in different stages of the CNN-based network to generate high-quality CAM in the model inference. Specifically, the semantic structure modeling module (SSM) is first proposed to generate the class-agnostic semantic correlation representation, where each item denotes the affinity degree between one category of objects and all the others. Then the structured feature representation is explored to polish an immature CAM via the dot product operation. Finally, the polished CAMs from different backbone stages are fused as the output. The proposed method has the advantage of no parameters and does not need to be trained. Therefore, it can be applied to a wide range of weakly-supervised pixel-wise dense prediction tasks. Experimental results on both weakly-supervised object localization and weakly-supervised semantic segmentation tasks demonstrate the effectiveness of the proposed method, which achieves the new state-of-the-art results on these two tasks.