论文检索

Yan Zhu, Huan Ma, Changqing Zhang

With the development of Vision Foundation Models (VFMs) in recent years, Visual In-Context Learning (VICL) has become a better choice compared to modifying models in most scenarios. Different from retraining or fine-tuning models, VICL does not require modifications to the model's weights and architecture, and only needs a prompt with demonstrations to teach VFM how to solve tasks. Currently, significant computational cost for finding optimal prompts for every test sample hinders the deployment of VICL, as determining which demonstrations to use for constructing the prompt is very costly. In this paper, however, we find a counterintuitive phenomenon that most test samples actually achieve optimal performance under the same prompts, and searching for sample-level prompts only costs much time but results in completely identical prompts actually. Therefore, we propose task-level prompting to reduce the cost of searching for prompts during the inference stage and introduce two time-saving yet effective task-level prompt search strategies accordingly. Extensive experimental results show that our proposed method can identify near-optimal prompts and reach the best VICL performance with a minimal cost that prior work has never achieved.

Xuekang Zhu, Xiaochen Ma, Lei Su, Zhuohang Jiang, Bo Du, Xiwen Wang, Zeyu Lei, Wentao Feng, Chi-Man Pun, Ji-Zhe Zhou

The mesoscopic level serves as a bridge between the macroscopic and microscopic worlds, addressing gaps overlooked by both. Image manipulation localization (IML), a crucial technique to pursue truth from fake images, has long relied on low-level (microscopic-level) traces. However, in practice, most tampering aims to deceive the audience by altering image semantics. As a result, manipulation commonly occurs at the object level (macroscopic level), which is equally important as microscopic traces. Therefore, integrating these two levels into the mesoscopic level presents a new perspective for IML research. Inspired by this, our paper explores how to simultaneously construct mesoscopic representations of micro and macro information for IML and introduces the Mesorch architecture to orchestrate both. Specifically, this architecture i) combines Transformers and CNNs in parallel, with Transformers extracting macro information and CNNs capturing micro details, and ii) explores across different scales, assessing micro and macro information seamlessly. Additionally, based on the Mesorch architecture, the paper introduces two baseline models aimed at solving IML tasks through mesoscopic representation. Extensive experiments across four datasets have demonstrated that our models surpass the current state-of-the-art in terms of performance, computational complexity, and robustness.

Xingyu Zhu, Xiapu Luo, Xuetao Wei

Recent advancements in text-to-3D generation can generate neural radiance fields (NeRFs) with score distillation sampling, enabling 3D asset creation without real-world data capture. With the rapid advancement in NeRF generation quality, protecting the copyright of the generated NeRF has become increasingly important. While prior works can watermark NeRFs in a post-generation way, they suffer from two vulnerabilities. First, a delay lies between NeRF generation and watermarking because the secret message is embedded into the NeRF model post-generation through fine-tuning. Second, generating a non-watermarked NeRF as an intermediate creates a potential vulnerability for theft. To address both issues, we propose Dreamark to embed a secret message by backdooring the NeRF during NeRF generation. In detail, we first pre-train a watermark decoder. Then, Dreamark generates backdoored NeRFs in a way that the target secret message can be verified by the pre-trained watermark decoder on an arbitrary trigger viewport. We evaluate the generation quality and watermark robustness against image- and model-level attacks. Extensive experiments show that the watermarking process will not degrade the generation quality, and the watermark achieves 90+% accuracy among both image-level attacks (e.g., Gaussian noise) and model-level attacks (e.g., pruning attack).

Xingyu Zhu, Guanhui Ye, Chengdong Dong, Xiapu Luo, Shiyao Zhang, Xuetao Wei

Mesh watermark embeds secret messages in 3D meshes and decodes the message from watermarked meshes for ownership verification. Current watermarking methods directly hide secret messages in vertex and face sets of meshes. However, mesh is a discrete representation that uses vertex and face sets to describe a continuous signal, which can be discretized in other discrete representations with different vertex and face sets. This raises the question of whether the watermark can still be verified on the different discrete representations of the watermarked mesh. We conduct this research in an attack-then-defense manner by proposing a novel function space mesh watermark removal attack FuncEvade and then mitigating it through function space mesh watermarking FuncMark. In detail, FuncEvade generates a different discrete representation of a watermarked mesh by extracting it from the signed distance function of the watermarked mesh. We observe that the generated mesh can evade ALL previous watermarking methods. FuncMark mitigates FuncEvade by watermarking signed distance function through message-guided deformation. Such deformation can survive isosurfacing and thus be inherited by the extracted meshes for further watermark decoding. Extensive experiments demonstrate that FuncEvade achieves 100% evasion rate among all previous watermarking methods while achieving only 0.3% evasion rate on FuncMark. Besides, our FuncMark performs similarly on other metrics compared to state-of-the-art mesh watermarking methods.

Tianyi Zhu, Wei Shang, Dongwei Ren

Animation line inbetweening is a crucial step in animation production aimed at enhancing animation fluidity by predicting intermediate line arts between two key frames. However, existing methods face challenges in effectively addressing sparse pixels and significant motion in line art key frames. In literature, Chamfer Distance (CD) is commonly adopted for evaluating inbetweening performance. Despite achieving favorable CD values, existing methods often generate interpolated frames with line disconnections, especially for scenarios involving large motion. Motivated by this observation, we propose a simple yet effective interpolation method for animation line inbetweening that adopts thin-plate spline-based transformation to estimate coarse motion more accurately by modeling the keypoint correspondence between two key frames, particularly for large motion scenarios. Building upon the coarse estimation, a motion refine module is employed to further enhance motion details before final frame interpolation using a simple UNet model. Furthermore, to more accurately assess the performance of animation line inbetweening, we refine the CD metric and introduce a novel metric termed Weighted Chamfer Distance, which demonstrates a higher consistency with visual perception quality. Additionally, we incorporate Earth Mover's Distance and conduct user study to provide a more comprehensive evaluation. Our method outperforms existing approaches by delivering high-quality interpolation results with enhanced fluidity.

Minjie Zhu, Yichen Zhu, Ning Liu, Xin Liu, Zhiyuan Xu, Chaomin Shen, Yaxin Peng

Multimodal Large Language Models (MLLMs) have showcased impressive skills in tasks related to visual understanding and reasoning. Yet, their widespread application faces obstacles due to the high computational demands during both the training and inference phases, restricting their use to a limited audience within the research and user communities. In this paper, we investigate the design aspects of Multimodal Small Language Models (MSLMs) and propose an efficient multimodal assistant named Mipha, which is designed to create synergy among various aspects: visual representation, language models, and optimization strategies. We show that without increasing the volume of training data, our Mipha-3B outperforms the state-of-the-art large MLLMs, especially LLaVA-1.5-13B, on multiple benchmarks. Through detailed discussion, we provide insights and guidelines for developing strong MSLMs that rival the capabilities of MLLMs.

Lingting Zhu, Guying Lin, Jinnan Chen, Xinjie Zhang, Zhenchao Jin, Zhao Wang, Lequan Yu

While Implicit Neural Representations (INRs) have demonstrated significant success in image representation, they are often hindered by large training memory and slow decoding speed. Recently, Gaussian Splatting (GS) has emerged as a promising solution in 3D reconstruction due to its highquality novel view synthesis and rapid rendering capabilities, positioning it as a valuable tool for a broad spectrum of applications. In particular, a GS-based representation, 2DGS, has shown potential for image fitting. In our work, we present Large Images are Gaussians (LIG), which delves deeper into the application of 2DGS for image representations, addressing the challenge of fitting large images with 2DGS in the situation of numerous Gaussian points, through two distinct modifications: 1) we adopt a variant of representation and optimization strategy, facilitating the fitting of a large number of Gaussian points; 2) we propose a Level-of-Gaussian approach for reconstructing both coarse low-frequency initialization and fine high-frequency details. Consequently, we successfully represent large images as Gaussian points and achieve high-quality large image representation, demonstrating its efficacy across various types of large images.

Jiaying Zhu, Dong Li, Xueyang Fu, Gege Shi, Jie Xiao, Aiping Liu, Zheng-Jun Zha

The rise in sophisticated image forgery techniques, driven by advancements in image editing and generation, has posed new security challenges. Traditional methods, designed for specific tampering artifacts, struggle with out-of-distribution image forgery detection. In this paper, we propose a shift in paradigm, placing greater emphasis on the universal characteristics of authentic images, as opposed to solely focusing on specific forgery signals. We introduce an enhancement to the Masked Autoencoder (MAE), aptly termed the Forgery MAE (FMAE). This modification retains the inherent characteristics of natural images while integrating multi-source forgery information. Our implementation involves applying the lottery ticket hypothesis during pre-training to identify forgery-sensitive parameters, followed by their sparse fine-tuning to target the forgery detection and localization task. Concurrently, we develop a ``mixture of experts'' noise extractor to compile multi-source forgery data. Our FMAE effectively extracts forgery features and shows strong resilience against unseen forgeries. Extensive experiments across multiple datasets confirm our method's superior accuracy and generalization capability over existing techniques.

Jiawen Zhu, Huayi Tang, Xin Chen, Xinying Wang, Dong Wang, Huchuan Lu

Efficient tracking has garnered attention for its ability to operate on resource-constrained platforms for real-world deployment beyond desktop GPUs. Current efficient trackers mainly follow precision-oriented trackers, adopting a one-stream framework with lightweight modules. However, blindly adhering to the one-stream paradigm may not be optimal, as incorporating template computation in every frame leads to redundancy, and pervasive semantic interaction between template and search region places stress on edge devices. In this work, we propose a novel asymmetric Siamese tracker named AsymTrack for efficient tracking. AsymTrack disentangles template and search streams into separate branches, with template computing only once during initialization to generate modulation signals. Building on this architecture, we devise an efficient template modulation mechanism to unidirectional inject crucial cues into the search features, and design an object perception enhancement module that integrates abstract semantics and local details to overcome the limited representation in lightweight tracker. Extensive experiments demonstrate that AsymTrack offers superior speed-precision trade-offs across different platforms compared to the current state-of-the-arts. For instance, AsymTrack-T achieves 60.8% AUC on LaSOT and 224/81/84 FPS on GPU/CPU/AGX, surpassing HiT-Tiny by 6.0% AUC with higher speeds.

Jianhua Zhu, Wenqi Zhao, Yu Li, Xingjian Hu, Liangcai Gao

Handwritten Mathematical Expression Recognition (HMER) has extensive applications in automated grading and office automation. However, existing sequence-based decoding methods, which directly predict LaTeX sequences, struggle to understand and model the inherent tree structure of LaTeX and often fail to ensure syntactic correctness in the decoded results. To address these challenges, we propose a novel model named TAMER (Tree-Aware Transformer) for handwritten mathematical expression recognition. TAMER introduces an innovative Tree-aware Module while maintaining the flexibility and efficient training of Transformer. TAMER combines the advantages of both sequence decoding and tree decoding models by jointly optimizing sequence prediction and tree structure prediction tasks, which enhances the model's understanding and generalization of complex mathematical expression structures. During inference, TAMER employs a Tree Structure Prediction Scoring Mechanism to improve the structural validity of the generated LaTeX sequences. Experimental results on CROHME datasets demonstrate that TAMER outperforms traditional sequence decoding and tree decoding models, especially in handling complex mathematical structures, achieving state-of-the-art (SOTA) performance.

Haiyang Zhu, Xi Yang, Nannan Wang

Weakly supervised person search aims to detect and match individuals using only bounding box annotations jointly. The existing methods mainly alternate between the clustering stage and the training stage, where the former is responsible for instance level label allocation tasks and the latter needs to undertake proposal level label allocation tasks. In the clustering phase, the conventional use of the DBSCAN algorithm for clustering pedestrian instance features often neglects key contextual information such as scene context and relative positioning of individuals. During the training phase, the Region Proposal Network assigns labels based on the MaxIoU, which tends to produce locally ambiguous labels. Finally, the proposals updated to the memory bank with extensive background information tend to interfere with the task of pseudo-label generation. To address these issues, this paper proposes an Optimizing Label Assignment (OLA) for weakly supervised person search. Firstly, in the clustering phase, Context Aware Clustering is introduced to integrate contextual information and constraints, enhancing the accuracy of clustering. Secondly, in the training phase, we adopt Prototype Matching based on Optimal Transport theory to optimize label distribution from a global perspective. Furthermore, we propose Dual Memory Bank Enhancement that effectively enhances the accuracy of label assignment. Extensive experiments conducted on the CUHK-SYSU and PRW datasets demonstrate that our method achieves state-of-the-art performance in weakly supervised person search.

Guilin Zhu, Dongyue Wu, Changxin Gao, Runmin Wang, Weidong Yang, Nong Sang

Class incremental semantic segmentation (CISS) aims to segment new classes during continual steps while preventing the forgetting of old knowledge. Existing methods alleviate catastrophic forgetting by replaying distributions of previously learned classes using stored prototypes or features. However, they overlook a critical issue: in CISS, the representation of class knowledge is updated continuously through incremental learning, whereas prototype replay methods maintain fixed prototypes. This mismatch between updated representation and fixed prototypes limits the effectiveness of the prototype replay strategy. To address this issue, we propose the Adaptive prototype replay (Adapter) for CISS in this paper. Adapter comprises an adaptive deviation compensation (ADC) strategy and an uncertainty-aware constraint (UAC) loss. Specifically, the ADC strategy dynamically updates the stored prototypes based on the estimated representation shift distance to match the updated representation of old class. The UAC loss reduces prediction uncertainty, aggregating discriminative features to aid in generating compact prototypes. Additionally, we introduce a compensation-based prototype similarity discriminative (CPD) loss to ensure adequate differentiation between similar prototypes, thereby enhancing the efficiency of the adaptive prototype replay strategy. Extensive experiments on Pascal VOC and ADE20K datasets demonstrate that Adapter achieves state-of-the-art results and proves effective across various CISS tasks, particularly in challenging multi-step scenarios.

Chenyang Zhu, Kai Li, Yue Ma, Chunming He, Xiu Li

This paper introduces MultiBooth, a method that generates images from texts containing various concepts from users.Despite diffusion models bringing significant advancements for customized text-to-image generation, existing methods often struggle with multi-concept scenarios due to low concept fidelity and high inference cost. MultiBooth addresses these issues by dividing the multi-concept generation process into two phases: a single-concept learning phase and a multi-concept integration phase. During the single-concept learning phase, we employ a multi-modal image encoder and an efficient concept encoding technique to learn a concise and discriminative representation for each concept. In the multi-concept integration phase, we use bounding boxes to define the generation area for each concept within the cross-attention map. This method enables the creation of individual concepts within their specified regions, thereby facilitating the formation of multi-concept images. This strategy not only improves concept fidelity but also reduces additional inference cost. MultiBooth surpasses various baselines in both qualitative and quantitative evaluations, showcasing its superior performance and computational efficiency.

Ziqi Zhou, Weize Quan, Hailin Shi, Wei Li, Lili Wang, Dong-Ming Yan

Audio-driven talking head generation necessitates seamless integration of audio and visual data amidst the challenges posed by diverse input portraits and intricate correlations between audio and facial motions. In response, we propose a robust framework GoHD designed to produce highly realistic, expressive, and controllable portrait videos from any reference identity with any motion. GoHD innovates with three key modules: Firstly, an animation module utilizing latent navigation is introduced to improve the generalization ability across unseen input styles. This module achieves high disentanglement of motion and identity, and it also incorporates gaze orientation to rectify unnatural eye movements that were previously overlooked. Secondly, a conformer-structured conditional diffusion model is designed to guarantee head poses that are aware of prosody. Thirdly, to estimate lip-synchronized and realistic expressions from the input audio within limited training data, a two-stage training strategy is devised to decouple frequent and frame-wise lip motion distillation from the generation of other more temporally dependent but less audio-related motions, e.g., blinks and frowns. Extensive experiments validate GoHD's advanced generalization capabilities, demonstrating its effectiveness in generating realistic talking face results on arbitrary subjects.

Ziheng Zhou, Jinxing Zhou, Wei Qian, Shengeng Tang, Xiaojun Chang, Dan Guo

In the field of audio-visual learning, most research tasks focus exclusively on short videos. This paper focuses on the more practical Dense Audio-Visual Event Localization (DAVEL) task, advancing audio-visual scene understanding for longer, untrimmed videos. This task seeks to identify and temporally pinpoint all events simultaneously occurring in both audio and visual streams. Typically, each video encompasses dense events of multiple classes, which may overlap on the timeline, each exhibiting varied durations. Given these challenges, effectively exploiting the audio-visual relations and the temporal features encoded at various granularities becomes crucial. To address these challenges, we introduce a novel CCNet, comprising two core modules: the Cross-Modal Consistency Collaboration (CMCC) and the Multi-Temporal Granularity Collaboration (MTGC). Specifically, the CMCC module contains two branches: a cross-modal interaction branch and a temporal consistency-gated branch. The former branch facilitates the aggregation of consistent event semantics across modalities through the encoding of audio-visual relations, while the latter branch guides one modality's focus to pivotal event-relevant temporal areas as discerned in the other modality. The MTGC module includes a coarse-to-fine collaboration block and a fine-to-coarse collaboration block, providing bidirectional support among coarse- and fine-grained temporal features. Extensive experiments on the UnAV-100 dataset validate our module design, resulting in a new state-of-the-art performance in dense audio-visual event localization.

Zhao Zhou, Xiangcheng Du, Yingbin Zheng, Xingjiao Wu, Cheng Jin

This paper introduces a novel exemplar-based framework for reading Chinese texts in natural scene or document images. We present the Deep Exemplar-based Chinese Text Recognizer, which is structured to first identify candidate characters as exemplars from each text-line, and subsequently recognize them by retrieving analogous exemplars from a database. With text-line level annotations, we design the exemplar discovery network to simultaneously recognize texts and capture individual character positions in a weak-supervision manner. The exemplar retrieval module is then crafted to identify the most similar exemplar and propagate the corresponding character label. This enables us to effectively rectify the misrecognized characters and boost the performance of scene text recognition. Experiments on four scenarios of Chinese texts demonstrate the effectiveness of our proposed framework.

Zhao Zhou, Weizhong Zhang, Xiangcheng Du, Yingbin Zheng, Cheng Jin

Image-text matching is a crucial task that bridges visual and linguistic modalities. Recent research typically formulates it into the problem of maximizing the margin with the truly hardest negatives to enhance the learning efficiency and avoid the poor local optima. We argue that such formulation can lead to a serious limitation, i.e., under this formulation, conventional trainers would confine their horizon within the hardest negative examples, while other negative examples offer a range of semantic differences not present in the hardest negatives. In this paper, we propose an efficient negative distribution guided training framework for image-text matching to unlock the substantial promotion space left by the above limitation. Rather than simply incorporating additional negative examples into the training objective, which could diminish both the leading role of the hardest negatives in training and the effect of a large margin learning in producing a robust matching model, our central idea is to supply the objective with distributional information on the entire set of negative examples. To be precise, we first construct the sample similarity matrix based on several pretrained models to extract the distributional information of the entire negative sample dataset. Then we encode it into a margin regularization module to smooth the similarities differences of all negatives. This enhancement facilitates the capture of fine-grained semantic differences and guides the main learning process by maximizing the margin with hard negative examples. Furthermore, we propose a hardest negative rectification module to address the instability in hardest negative selection based on predicted similarity and to correct erroneous hardest negatives. We evaluate our method in combination with several state-of-the-art image-text matching methods, and our quantitative and qualitative experiments demonstrate its significant generalizability and effectiveness.

Zhao Zhou, Yiqun Wang, Weizhong Zhang, Yingbin Zheng, Xiangcheng Du, Cheng Jin

Model ensembling is a widely used technique that enhances performance in image-text matching tasks by combining multiple models, each trained with different initializations. However, the inefficiencies associated with training several models and generating outputs from them constrain their practical applicability. In this paper, we argue that while the parameters of two randomly initialized models can differ significantly, their feature distributions can be similar at certain stages. By employing a proposed technique called cross-modal realignment, we demonstrate that features derived from differently initialized models maintain similarity at the feature extraction stage and can be effectively transformed by fine-tuning a small number of parameters. These findings provide an efficient way to achieve ensemble-like performance within a single model. Specifically, we propose a Feature Diversification Framework (FDF) that emulates the outputs of multiple model initializations to generate diverse features from a common shared feature. Firstly, we introduce feature conversion methods to transform shared features into a set of distinct features. Next, a realignment training strategy is presented to optimize negative pairs for realigning these transformed features, thereby enhancing their diversification to resemble the outputs of different models. Additionally, we propose a reweighting module that assigns weights to these features, enabling a weighted fusion approach for robust feature representation. Extensive experiments on the Flickr30K and MS-COCO datasets demonstrate the effectiveness and generalizability of our framework.

Yuyan Zhou, Ye Li, Lei Feng, Sheng-Jun Huang

Recent studies showed that the generalization of neural networks is correlated with the sharpness of the loss landscape and flat minima suggests a better generalization ability than sharp minima. In this paper, we propose a novel method called optimum shifting, which changes the parameters of a neural network from a sharp minimum to a flatter one while maintaining the same training loss value. Our method is based on the observation that when the input and output of a neural network are fixed, the matrix multiplications within the network can be treated as systems of under-determined linear equations, enabling adjustment of parameters in the solution space, which can be simply accomplished by solving a constrained optimization problem. Furthermore, we introduce a practical stochastic optimum shifting technique utilizing the neural collapse theory to reduce computational costs and provide more degrees of freedom for optimum shifting. Extensive experiments with various deep neural network architectures on benchmark datasets demonstrate the effectiveness of our method.

Yunshuai Zhou, Junbo Qiao, Jincheng Liao, Wei Li, Simiao Li, Jiao Xie, Yunhang Shen, Jie Hu, Shaohui Lin

Knowledge distillation (KD) is a valuable yet challenging approach that enhances a compact student network by learning from a high-performance but cumbersome teacher model. However, previous KD methods for image restoration overlook the state of the student during the distillation, adopting a fixed solution space that limits the capability of KD. Additionally, relying solely on L1-type loss struggles to leverage the distribution information of images. In this work, we propose a novel dynamic contrastive knowledge distillation (DCKD) framework for image restoration. Specifically, we introduce dynamic contrastive regularization to perceive the student's learning state and dynamically adjust the distilled solution space using contrastive learning. Additionally, we also propose a distribution mapping module to extract and align the pixel-level category distribution of the teacher and student models. Note that the proposed DCKD is a structure-agnostic distillation framework, which can adapt to different backbones and can be combined with methods that optimize upper-bound constraints to further enhance model performance. Extensive experiments demonstrate that DCKD significantly outperforms the state-of-the-art KD methods across various image restoration tasks and backbones.