论文检索

Chengjie Ge, Xueyang Fu, Peng He, Kunyu Wang, Chengzhi Cao, Zheng-Jun Zha

Leveraging its robust linear global modeling capability, Mamba has notably excelled in computer vision. Despite its success, existing Mamba-based vision models have overlooked the nuances of event-driven tasks, especially in video reconstruction. Event-based video reconstruction (EBVR) demands spatial translation invariance and close attention to local event relationships in the spatio-temporal domain. Unfortunately, conventional Mamba algorithms apply static window partitions and standard reshape scanning methods, leading to significant losses in local connectivity. To overcome these limitations, we introduce EventMamba—a specialized model designed for EBVR task. EventMamba innovates by incorporating random window offset (RWO) in the spatial domain, moving away from the restrictive fixed partitioning. Additionally, it features a new consistent traversal serialization approach in the spatio-temporal domain, which maintains the proximity of adjacent events both spatially and temporally. These enhancements enable EventMamba to retain Mamba’s robust modeling capabilities while significantly preserving the spatio-temporal locality of event data. Comprehensive testing on multiple datasets shows that EventMamba markedly enhances video reconstruction, drastically improving computation speed while delivering superior visual quality compared to Transformer-based methods.

Xianqiang Gao, Pingrui Zhang, Delin Qu, Dong Wang, Zhigang Wang, Yan Ding, Bin Zhao

3D Object Affordance Grounding aims to predict the functional regions on a 3D object and has laid the foundation for a wide range of applications in robotics. Recent advances tackle this problem via learning a mapping between 3D regions and a single human-object interaction image. However, the geometric structure of the 3D object and the object in the human-object interaction image are not always consistent, leading to poor generalization. To address this issue, we propose to learn generalizable invariant affordance knowledge from multiple human-object interaction images within the same affordance category. Specifically, we introduce the Multi-Image Guided Invariant-Feature-Aware 3D Affordance Grounding (MIFAG) framework. It grounds 3D object affordance regions by identifying common interaction patterns across multiple human-object interaction images. First, the Invariant Affordance Knowledge Extraction Module (IAM) utilizes an iterative updating strategy to gradually extract aligned affordance knowledge from multiple images and integrate it into an affordance dictionary. Then, the Affordance Dictionary Adaptive Fusion Module (ADM) learns comprehensive point cloud representations that consider all affordance candidates in multiple images. Besides, the Multi-Image and Point Affordance (MIPA) benchmark is constructed and our method outperforms existing state-of-the-art methods on various experimental comparisons.

Mingze Gao, Jingyu Liu, Mingda Li, Jiangtao Xie, Qingbin Liu, Kevin Zhao, Xi Chen, Hui Xiong

Multimodal Large Language Models (MLLMs) have significantly improved performance across various image-language applications. Recently, there has been a growing interest in adapting image pre-trained MLLMs for video-related tasks. However, most efforts concentrate on enhancing the vision encoder and projector components, while the core part, Large Language Models (LLMs), remains comparatively under-explored. In this paper, we propose two strategies to enhance the model's capability in video understanding tasks by improving inter-layer attention computation in LLMs. Specifically, the first approach focuses on the enhancement of Rotary Position Embedding (RoPE) with Temporal-Aware Dual RoPE, which introduces temporal position information to strengthen the MLLM's temporal modeling capabilities while preserving the relative position relationships of both visual and text tokens. The second approach involves enhancing the Attention Mask with the Frame-wise Block Causal Attention Mask, a simple yet effective method that broadens visual token interactions within and across video frames while maintaining the causal inference mechanism. Based on these proposed methods, we adapt LLaVA for video understanding tasks, naming it Temporal-Considered LLaVA (TC-LLaVA). Our TC-LLaVA achieves new state-of-the-art performance across various video understanding benchmarks with only supervised fine-tuning (SFT) on video-related datasets.

Jun Gao, Qian Qiao, Tianxiang Wu, Zili Wang, Ziqiang Cao, Wenjie Li

In-context learning (ICL) advances Large Language Models (LLMs) exhibiting emergent ability on downstream tasks without updating billions of parameters. However, in the area of multimodal Large Language Models (MLLMs), two problems hinder the application of multimodal ICL: (1) Most primary MLLMs are only trained on single-image datasets, making them unable to read extra multimodal demonstrations. (2) With the demonstrations increasing, thousands of visual tokens highly challenge hardware and degrade ICL performance. During preliminary explorations, we discovered that the inner LLM focuses more on the linguistic modality within multimodal demonstrations during generation. Therefore, we propose a general and lightweight framework AIM to tackle the mentioned problems through Aggregating Image information of Multimodal demonstrations to the latent space of the corresponding textual labels. After aggregation, AIM substitutes each demonstration with generated fused virtual tokens whose length is reduced to the same as its texts. Except for shortening input length, AIM further upgrades MLLMs pre-trained on image-text pairs to support multimodal ICL, as images from demonstrations are disregarded. Furthermore, benefiting from aggregating different demonstrations independently, AIM configures Demonstration Bank (DB) to avoid repeated aggregation, which significantly boosts model efficiency. We build AIM upon QWen-VL and LLaVA-Next, and AIM is comprehensively evaluated on image caption, VQA, and hateful speech detection. Outstanding results reveal that AIM provides an efficient and effective solution in upgrading MLLMs for multimodal ICL.

Ge Gao, Ho Man Kwan, Fan Zhang, David Bull

Neural video compression has recently demonstrated significant potential to compete with conventional video codecs in terms of rate-quality performance. These learned video codecs are however associated with various issues related to decoding complexity (for autoencoder-based methods) and/or system delays (for implicit neural representation (INR) based models), which currently prevent them from being deployed in practical applications. In this paper, targeting a practical neural video codec, we propose a novel INR-based coding framework, PNVC, which innovatively combines autoencoder-based and overfitted solutions. Our approach benefits from several design innovations, including a new structural reparameterization-based architecture, hierarchical quality control, modulation-based entropy modeling, and scale-aware positional embedding. Supporting both low delay (LD) and random access (RA) configurations, PNVC outperforms existing INR-based codecs, achieving nearly 35%+ BD-rate savings against HEVC HM 18.0 (LD) - almost 10% more compared to one of the state-of-the-art INR-based codecs, HiNeRV and 5% more over VTM 20.0 (LD), while maintaining 20+ FPS decoding speeds for 1080p content. This represents an important step forward for INR-based video coding, moving it towards practical deployment.

Lianqiang Gan, Junyu Lai, Jingze Ju, Lianli Gao, Yi Bin

Videos inherently contain complex temporal dynamics across various spatial directions, often entangled in ways that obscure effective dynamic extraction. Previous studies typically process video spatiotemporal features without disentangling, which hampers their ability to extract dynamic information. Additionally, the extraction of dynamics is disrupted by transient high-dynamic information in video sequences, e.g., noise or flicker, which has received limited attention in the literature. To tackle those problems, this paper proposes the Disentangling and Filtering Dynamics Network (DFDNet). Firstly, to disentangle the interwoven dynamics, DFDNet decomposes the spatially encoded video sequences into lower dimensional sequences. Secondly, a learnable threshold filter is proposed to eliminate the transient high-dynamic information. Thirdly, the model incorporates an MLP to extract the temporal dependencies from the disentangled and filtered sequences. DFDNet demonstrates competitive performance across four chosen datasets, including both low and high-resolution videos. Specifically, on the low-resolution Moving MNIST dataset, DFDNet achieves a 19% improvement on MSE over the previous state-of-the-art model. On the high-resolution SJTU4K dataset, it outperforms the previous state-of-the-art model by 10% on the LPIPS metric under similar inference time.

Keke Gai, Dongjue Wang, Jing Yu, Mohan Wang, Liehuang Zhu, Qi Wu

Multi-modal Federated Learning (MFL) is a distributed machine learning paradigm that enables multiple participants with multi-modal data to collaboratively train a global model for multi-modal tasks without sharing their local data. MFL typically deploys the trained global model as an Embedding-as-a-Service (EaaS), allowing participants to obtain embeddings for downstream tasks. However, it increases the risk of unauthorized copying and leakage of the model. Protecting the ownership of the MFL model while maintaining model performance is challenging. In this paper, we propose the first general model ownership protection framework for MFL, named MFL-Owner. MFL-Owner decouples the watermarking process from the model training process and addresses both ownership verification and traceability, effectively safeguarding the interests of the MFL collective. MFL-Owner leverages the concept of orthogonal transformations by incorporating a linear transformation matrix with orthogonal constraints into the model, achieving high-quality ownership verification and traceability with minimal impact on model performance. To enhance the practicality of the watermark and prevent conflicts among multiple clients during tracing, we propose a trigger dataset selection method based on out-of-distribution data combined with Gaussian noise perturbation. Our experiments on multiple datasets demonstrate that MFL-Owner is effective for model ownership verification and traceability for MFL.

Xinghe Fu, Zhiyuan Yan, Taiping Yao, Shen Chen, Xi Li

The generalization problem is broadly recognized as a critical challenge in detecting deepfakes. Most previous work believes that the generalization gap is caused by the differences among various forgery methods. However, our investigation reveals that the generalization issue can still occur when forgery-irrelevant factors shift. In this work, we identify two biases that detectors may also be prone to overfitting: position bias and content bias, as depicted in Fig. 1. For the position bias, we observe that detectors are prone to “lazily” depending on the specific positions within an image (e.g., central regions even no forgery). As for content bias, we argue that detectors may potentially and mistakenly utilize forgery-unrelated information for detection (e.g., background, and hair). To intervene in these biases, we propose two branches for shuffling and mixing with tokens in the latent space of transformers. For the shuffling branch, we rearrange the tokens and corresponding position embedding for each image while maintaining the local correlation. For the mixing branch, we randomly select and mix the tokens in the latent space between two images with the same label within the mini-batch to recombine the content information. During the learning process, we align the outputs of detectors from different branches in both feature space and logit space. Contrastive losses for features and divergence losses for logits are applied to obtain unbiased feature representation and classifiers. We demonstrate and verify the effectiveness of our method through extensive experiments on widely used evaluation datasets.

Teng Fu, Haiyang Yu, Ke Niu, Bin Li, Xiangyang Xue

Multiple Object Tracking (MOT) is a fundamental task in computer vision. Existing methods utilize motion information or appearance information to perform object tracking. However, these algorithms still struggle with special circumstances, such as occlusion and blurring in complex scenes. Inspired by the fact that people can pinpoint objects through verbal descriptions, we explore performing long-term robust tracking using semantic features of objects. Motivated by the success of the multimodal foundation model in text-image alignment, we reconsider the appearance feature extraction module in MOT and propose a Foundation model Driven multi-object tracker (FDTracker). Specifically, we propose a two-stage trained appearance feature extractor. In the first stage, using a single image of the object as input, the model could capture the attributes of objects with the assistance of natural language instructions. In the second stage, using a sequence of images of objects as input, the model learns how to use these attributes to distinguish between different objects and connect the same object at different times. Finally, for coordinating appearance and motion information, we propose a reasonable combined strategy, which better facilitates trajectory assignment and reconnection. Extensive experiments on benchmarks demonstrate the robustness of FDTracker.

Chenlin Fu, Yingying Zhu

Oriented object detection is crucial for complex scenes such as aerial images and industrial inspection, providing precise delineation by minimizing background interference. Recently, the weakly-supervised detector paradigm H2RBox has demonstrated promise in learning rotated bounding box (RBox) from the more readily available horizontal bounding box (HBox), alleviating the scarcity and high cost of RBox annotations. However, these H2RBox-based methods have primarily focused on the gap in orientation information between HBox- and RBox-supervised approaches, overlooking the gap in training sample selection. In response, we propose the Adaptive Fine-grained Sample Mining (AFSM) strategy, which improves the selection of fine-grained training samples in HBox-supervised methods. AFSM assigns the best-matching prediction RBox to each ground truth (GT) HBox and selects positive samples based on these paired boxes. Furthermore, to effectively filter the best-matching prediction RBox for AFSM, we introduce the Prediction Rbox Assignment (PRA) scheme, employing Kullback-Leibler Divergence (KLD) as a localization quality metric. Additionally, we introduce an improved self-supervised branch loss (Lss) to address the symmetry of weakly-supervised branch prediction boxes. Incorporating these core components (AFSM, PRA, and Lss), we develop an end-to-end network architecture (BGHR) to further bridge the gap between HBox- and RBox-supervised oriented object detection. Extensive experiments on DOTA-v1.0 and DIOR-R demonstrate that BGHR achieves state-of-the-art performance compared to HBox-supervised methods without additional overhead. Even when benchmarked against fully supervised FCOS, our method still exhibits a slight performance advantage.

Zhida Feng, Li Chen, Yuenan Sun, Jiaxiang Liu, Shikun Feng

ControlNet has significantly advanced controllable image generation by integrating dense conditions (such as depth and canny edges) with text-to-image diffusion models. However, ControlNet's integration requires an additional amount nearly equal to half of the base diffusion model's parameters, making it inefficient. To address this, we introduce Simple-ControlNet, an efficient and streamlined network for controllable text-to-image generation. It employs a single-scale projection layer to incorporate condition information into the denoising U-Net. It is supplemented by Low-Rank Adapter (LoRA) parameters to facilitate condition learning. Impressively, Simple-ControlNet requires fewer than 3 million parameters for the control mechanism, substantially less than the 300 million needed by ControlNet. Our extensive experiments confirm that Simple-ControlNet matches and surpasses ControlNet's performance across a broad range of tasks and base diffusion models, showcasing its utility and efficiency.

Yi Feng, Yu Han, Xijing Zhang, Tanghui Li, Yanting Zhang, Rui Fan

Inferring the 3D structure of a scene from a single image is an ill-posed and challenging problem in the field of vision-centric autonomous driving. Existing methods usually employ neural radiance fields to produce voxelized 3D occupancy, lacking instance-level semantic reasoning and temporal photometric consistency. In this paper, we propose ViPOcc, which leverages the visual priors from vision foundation models (VFMs) for fine-grained 3D occupancy prediction. Unlike previous works that solely employ volume rendering for RGB and depth image reconstruction, we introduce a metric depth estimation branch, in which an inverse depth alignment module is proposed to bridge the domain gap in depth distribution between VFM predictions and the ground truth. The recovered metric depth is then utilized in temporal photometric alignment and spatial geometric alignment to ensure accurate and consistent 3D occupancy prediction. Additionally, we also propose a semantic-guided non-overlapping Gaussian mixture sampler for efficient, instance-aware ray sampling, which addresses the redundant and imbalanced sampling issue that still exists in previous state-of-the-art methods. Extensive experiments demonstrate the superior performance of ViPOcc in both 3D occupancy prediction and depth estimation tasks on diverse public datasets.

Tonghui Feng, Chunsheng Yan, Qianru Wang, Jiangtao Cui, Xiaotian Qiao

Visual text generation, which aims to generate photo-realistic images with coherent and well-formed scene text being rendered, has attracted widespread attention. Although recent works have achieved promising performance, the limited flexibility and controllability hinder their practical applications. We observe that different from natural objects, visual text in real scenes often has an arbitrarily shaped structure with different granularities (i.e., character, word, or line). In this paper, we consider the modality gap between image and text, and propose a new separation and composition pipeline for flexible and controllable visual text generation from only text prompts. At the core of our framework is a novel Hierarchical and Directional Layout representation, i.e., HDLayout, which can model the sequential and multi-granularity nature of the visual text. Under this formulation, we are able to generate arbitrarily shaped visual text automatically. Extensive experiments demonstrate that our method outperforms several strong baselines in a variety of scenarios both qualitatively and quantitatively, yielding state-of-the-art performances on arbitrarily shaped visual text generation.

Siyang Feng, Huadeng Wang, Chu Han, Zhenbing Liu, Hualong Zhang, Rushi Lan, Xipeng Pan

Image-level weakly supervised semantic segmentation (WSSS) reduces the dependence on high-quality data annotation, which plays a crucial role in computational pathology. Benefit from the ability to localize the objects with only binary labels, Class Activation Map (CAM) is a widely used method to initial pseudo masks. However, due to the low contrast among different tissues in histopathological images, most existing CAM-based methods perform poorly in gland segmentation. We retrospect this process and find that class consistency and semantic consistency can guide the network to effectively distinguish confusing pixels and generate fine-grained pseudo masks. Specifically, for class consistency, we propose Consistency Correlation Attention (CCA) to encourage the network to focus on the contribution of class features to semantic dependencies. For semantic consistency, we propose Multi-scale Pyramid Fusion Pooling (MPFP) to aggregate coarse-to-fine global semantic information from CAMs at multiple spatial resolutions, thus identifying class localization. Additionally, we introduce a Purified Labels Filtration (PLF) strategy during the segmentation phase to mitigate the noisy supervision signal and improve the segmentation quality of the model. Extensive experiments show that the our method achieves new state-of-the-art results on three publicly available gland datasets. Furthermore, our method demonstrates impressive domain adaptation capability, achieving satisfactory results with only a small portion of samples when faced with unseen domain data.

Mingtao Feng, Fenghao Tian, Jianqiao Luo, Zijie Wu, Weisheng Dong, Yaonan Wang, Ajmal Saeed Mian

Visual cross view geo-localization is generally approached within a joint retrieval-and-calibration framework. However, existing methods overlook semantic ambiguities arising from query and reference images characterized by low overlap, dynamic foregrounds, viewpoint changes, and perceptual aliasing. This makes it challenging to automatically control the relative importance of the two tasks, potentially compromising the retrieval task in favor of the offset regression. Consequently, the model may encounter conflicting dominating gradients during joint training. To address this, we propose to model the semantic ambiguity during the offset regression process by integrating associated uncertainty scores, represented as 2D Gaussian distributions, to mitigate negative transfer effects within the joint tasks. We further introduce an uncertainty-aware similarity metric to enhance similarity assessment between query and reference images, accounting for their semantic ambiguities. This metric propagates uncertainty scores into the retrieval task, focusing on certain samples and learning discriminative feature embeddings, allowing the model to adaptively handle conflicting dominating gradients during joint training. Extensive experiments demonstrate that our method improves the overall performance of the joint tasks, achieving state-of-the-art results on the VIGOR and CVACT datasets.

Kunyu Feng, Yue Ma, Bingyuan Wang, Chenyang Qi, Haozhe Chen, Qifeng Chen, Zeyu Wang

Despite recent advances in UNet-based image editing, methods for shape-aware object editing in high-resolution images are still lacking. Compared to UNet, Diffusion Transformers (DiT) demonstrate superior capabilities to effectively capture the long-range dependencies among patches, leading to higher-quality image generation. In this paper, we propose DiT4Edit, the first Diffusion Transformer-based image editing framework. Specifically, DiT4Edit uses the DPM-Solver inversion algorithm to obtain the inverted latents, reducing the number of steps compared to the DDIM inversion algorithm commonly used in UNet-based frameworks. Additionally, we design unified attention control and patch merging, tailored for transformer computation streams. This integration allows our framework to generate higher-quality edited images faster. Our design leverages the advantages of DiT, enabling it to surpass UNet structures in image editing, especially in high-resolution and arbitrary-size images. Extensive experiments demonstrate the strong performance of DiT4Edit in various editing scenarios, highlighting the potential of diffusion transformers for image editing.

Haoxuan Feng, Haohui Zhou, Tian Ye, Sixiang Chen, Lei Zhu

Defocus deblurring is a challenging task due to the spatially varying nature of defocus blur with multiple plausible solutions of a single given image. However, most existing methods falter when faced with extensive and variable defocus blur, either ignoring it or relying on additional loss functions to enhance perceptual quality. This often results in unrealistic reconstructions and compromised generalizability. In this paper, we propose a novel Residual Diffusion Deblurring Model framework for single image defocus deblurring. Our approach integrates a pre-trained defocus map estimator and a lightweight pre-deblur module with a learnable receptive field, providing crucial posterior information to effectively address large-scale and varying shaped defocus blur. In addition, a carefully-design denoising network enables the generation of diverse reconstructions from a single input. This approach not only significantly improves the perceptual quality of defocus deblurring outputs through multi-step residual learning, but also offers a more efficient inference strategy. Experimental results demonstrate that our method achieves competitive performance on real-world defocus deblurring image datasets across both perceptual and distortion evaluation metrics.

Dong Feng, Ping Guo, Encheng Peng, Mingmin Zhu, Wenhao Yu, Peng Wang

Manipulating human poses based on natural language is an emerging research field that has traditionally focused on coarse commands such as “walking” or “dancing.” However, fine-grained pose manipulation, like instructing “put both hands in front of the stomach,” remains underexplored. In this paper, we introduce PoseLLaVA, a pioneering model that integrates SMPL-based pose representations into the multimodal LLaVA framework. Through a novel pose encoder decoder mechanism, PoseLLaVA achieves precise alignment between pose, textual, and visual modalities, enabling detailed control over pose manipulation tasks. PoseLLaVA excels in three key tasks: pose estimation, generation, and adjustment, all driven by detailed language instructions. We further introduce a fine-grained pose adjustment dataset PosePart, where each sample contains an initial pose and a target pose, along with specific instructions for adjustments, mimicking the guidance a human instructor might provide. Extensive evaluations across these tasks demonstrate significant improvements over existing methods, including metrics such as MPJPE and PA-MPJPE, which measure SMPL reconstruction errors, and Recall rates, which assess feature alignment across modalities. Specifically, PoseLLaVA reduces MPJPE errors by more than 20% compared to state-of-the-art methods in pose adjustment and generation tasks. Additionally, we demonstrate the feasibility of combining PoseLLaVA with generative models, such as diffusion, for pose image editing, highlighting its potential applications in language-controlled pose manipulation.

Chun-Mei Feng, Yang Bai, Tao Luo, Zhen Li, Salman Khan, Wangmeng Zuo, Rick Siow Mong Goh, Yong Liu

Albeit progress has been made in Composed Image Retrieval (CIR), we empirically find that a certain percentage of failure retrieval results are not consistent with their relative captions. To address this issue, this work provides a Visual Question Answering (VQA) perspective to boost the performance of CIR. The resulting VQA4CIR is a post-processing approach and can be directly plugged into existing CIR methods. Given the top-C retrieved images by a CIR method, VQA4CIR aims to decrease the adverse effect of the failure retrieval results being inconsistent with the relative caption. To find the retrieved images inconsistent with the relative caption, we resort to the "QA generation → VQA" self-verification pipeline. For QA generation, we suggest fine-tuning LLM (e.g., LLaMA) to generate several pairs of questions and answers from each relative caption. We then fine-tune LVLM (e.g., LLaVA) to obtain the VQA model. By feeding the retrieved image and question to the VQA model, one can find the images inconsistent with relative caption when the answer by VQA is inconsistent with the answer in the QA pair. Consequently, the CIR performance can be boosted by modifying the ranks of inconsistently retrieved images. Experimental results show that our proposed method outperforms state-of-the-art CIR methods on the CIRR and Fashion-IQ datasets.

Chen Feng, Ziquan Liu, Zhuo Zhi, Ilija Bogunovic, Carsten Gerner-Beuerle, Miguel Rodrigues

It is widely known that state-of-the-art machine learning models, including vision and language models, can be seriously compromised by adversarial perturbations. It is therefore increasingly relevant to develop capabilities to certify their performance in the presence of the most effective adversarial attacks. Our paper offers a new approach to certify the performance of machine learning models in the presence of adversarial attacks with population level risk guarantees. In particular, we introduce the notion of (α,ζ)-safe machine learning model. We propose a hypothesis testing procedure, based on the availability of a calibration set, to derive statistical guarantees providing that the probability of declaring that the adversarial (population) risk of a machine learning model is less than α (i.e. the model is safe), while the model is in fact unsafe (i.e. the model adversarial population risk is higher than α), is less than ζ. We also propose Bayesian optimization algorithms to determine efficiently whether a machine learning model is (α,ζ)-safe in the presence of an adversarial attack, along with statistical guarantees. We apply our framework to a range of machine learning models - including various sizes of vision Transformer (ViT) and ResNet models - impaired by a variety of adversarial attacks, such as PGDAttack, MomentumAttack, GenAttack and BanditAttack, to illustrate the operation of our approach. Importantly, we show that ViT's are generally more robust to adversarial attacks than ResNets, and large models are generally more robust than smaller models. Our approach goes beyond existing empirical adversarial risk-based certification guarantees. It formulates rigorous (and provable) performance guarantees that can be used to satisfy regulatory requirements mandating the use of state-of-the-art technical tools.