The malicious use and widespread dissemination of deepfake pose a significant crisis of trust. Current deepfake detection models can generally recognize forgery images by training on a large dataset. However, the accuracy of detection models degrades significantly on images generated by new deepfake methods due to the difference in data distribution. To tackle this issue, we present a novel incremental learning framework that improves the generalization of deepfake detection models by continual learning from a small number of new samples. To cope with different data distributions, we propose to learn a domain-invariant representation based on supervised contrastive learning, preventing overfit to the insufficient new data. To mitigate catastrophic forgetting, we regularize our model in both feature-level and label-level based on a multi-perspective knowledge distillation approach. Finally, we propose to select both central and hard representative samples to update the replay set, which is beneficial for both domain-invariant representation learning and rehearsal-based knowledge preserving. We conduct extensive experiments on four benchmark datasets, obtaining the new state-of-the-art average forgetting rate of 7.01 and average accuracy of 85.49 on FF++, DFDC-P, DFD, and CDF2. Our code is released at \textcolorblue https://github.com/DeepFakeIL/DFIL.
论文检索
输入标题、作者或关键词，从 12,319 篇学术成果中精准定位
Currently, reference-based super-resolution (RefSR) techniques leverage high-resolution (HR) reference images to provide useful content and texture information for low-resolution (LR) images during the super-resolution (SR) process. Nevertheless, it is time-consuming, laborious, and even impossible in some cases to find high-quality reference images. To tackle this problem, we propose a brand-new self-reference image super-resolution approach using a pre-trained diffusion large model and a window adjustable transformer, termed DWTrans. Our proposed method does not require explicitly inputting manually acquired reference images during training and inference. Specifically, we feed the degraded LR images into a pre-trained stable diffusion large model to automatically generate corresponding high-quality self-reference (SRef) images that provide valuable high-frequency details for the LR images in the process of SR. To extract valuable high-frequency information in SRef images, we design a window adjustable transformer with both non-adjustable window layer (NWL) and adjustable window layer (AWL). The NWL learns local features from LR images using a dense window, while the AWL acquires global features from the SRef images using a random sparse window. Furthermore, to fully utilize the high-frequency features in the SRef image, we introduce the adaptive deformable fusion module to adaptively fuse the features of the LR and SRef images. Experimental results validate that our proposed DWTrans outperforms state-of-the-art methods on various benchmark datasets both quantitatively and visually.
AI has made significant progress in solving math problems, but geometry problems remain challenging due to their reliance on both text and diagrams. In the text description, symbolic characters such as "ABC" often serve as a bridge to connect the corresponding diagram. However, by simply tokenizing symbolic characters into individual letters (e.g., 'A', 'B' and 'C'), existing works fail to study them explicitly and thus lose the semantic relationship with the diagram. In this paper, we develop a symbolic character-aware model to fully explore the role of these characters in both text and diagram understanding and optimize the model under a multi-modal reasoning framework. In the text encoder, we propose merging individual symbolic characters to form one semantic unit along with geometric information from the corresponding diagram. For the diagram encoder, we pre-train it under a multi-label classification framework with the symbolic characters as labels. In addition, we enhance the geometry diagram understanding ability via a self-supervised learning method under the masked image modeling auxiliary task. By integrating the proposed model into a general encoder-decoder pipeline for solving geometry problems, we demonstrate its superiority on two benchmark datasets, including GeoQA and Geometry3K, with extensive experiments. Specifically, on GeoQA, the question-solving accuracy is increased from 60.0% to 64.1%, achieving a new state-of-the-art accuracy; on Geometry3K, we reduce the question average solving steps from 6.9 down to 6.0 with marginally higher solving accuracy.
Fine-grained image classification focuses on distinguishing objects from different similar subcategories, which requires the classification model to extract subtle yet discriminative descriptors. Recent Vision Transformer (ViT) has shown an enormous potential for this challenging task, but previous ViT-based methods have primarily focused on improving the relationship between image patches, neglecting the limited expressive capability caused by the single class token.To address this limitation, we propose to learn a Multi-part Token Transformer (MpT-Trans), which extends the class token to multiple tokens presenting various parts, enhancing the model's capability of extracting discriminative information. Specifically, our MpT-Trans model interpolates the vision transformer framework with two modules: (i) the Part-wise Shift Learning (PwSL) module is proposed to extend the single class token to a set of part tokens with differentiable shifts, enabling the model to extract informative representations from different perspectives; (ii) the Dual Contrastive Learning (DuCL) module is introduced to exploit the inter-class and inter-part relationships to regularize the learning of part tokens, enhancing their diversity and discrimination for accurate classification. Extensive experiments and ablation study demonstrate that the proposed MpT-Trans achieves state-of-the-art performance on various fine-grained image benchmark datasets, demonstrating the effectiveness of our proposed method.
Source-Free domain adaptive Semantic Segmentation (SFSS) aims to transfer knowledge from source domain to the target domain with only pre-trained source segmentation model and the unlabeled target dataset. Only a few works have been researched for SFSS, relying on entropy minimization, pseudo-labeling. Nevertheless, due to the domain bias, these methods tend to suffering from the confusion of classes with a similar visual appearance in different domains. To address the above issue, we propose to enhance discriminability towards target samples with masked image modeling to model spatial context relations as additional recognition clues. Specifically, we design a novel Dual-Level Masked Consistency method, which explicitly encourages the model to learn comprehensive context relations, i.e. patch-wise context and channel-wise context, on the target domain. By randomly masking target images and forcing the model to reconstruct predictions of the entire image with left unmasked part, the model has to make full use of spatially contextual information. To take a step further, we propose a novel masking strategy considering both local context and global context information by applying patch-wise masking on image patches and channel-wise masking on latent features. Notably, patch-wise context learning and channel-wise context learning can complement each other. Extensive experiments demonstrate the effectiveness of our proposed method and our method achieves state-of-the-art performance on two synthetic-to-real benchmarks: GTA5→Cityscapes and SYNTHIA→Cityscapes.
The audio-visual segmentation (AVS) task aims to segment sounding objects from a given video. Existing works mainly focus on fusing audio and visual features of a given video to achieve sounding object masks. However, we observed that prior arts are prone to segment a certain salient object in a video regardless of the audio information. This is because sounding objects are often the most salient ones in the AVS dataset. Thus, current AVS methods might fail to localize genuine sounding objects due to the dataset bias. In this work, we present an audio-visual instance-aware segmentation approach to overcome the dataset bias. In a nutshell, our method first localizes potential sounding objects in a video by an object segmentation network, and then associates the sounding object candidates with the given audio. We notice that an object could be a sounding object in one video but a silent one in another video. This would bring ambiguity in training our object segmentation network as only sounding objects have corresponding segmentation masks. We thus propose a silent object-aware segmentation objective to alleviate the ambiguity. Moreover, since the category information of audio is unknown, especially for multiple sounding sources, we propose to explore the audio-visual semantic correlation and then associate audio with potential objects. Specifically, we attend predicted audio category scores to potential instance masks and these scores will highlight corresponding sounding instances while suppressing inaudible ones. When we enforce the attended instance masks to resemble the ground-truth mask, we are able to establish audio-visual semantics correlation. Experimental results on the AVS benchmarks demonstrate that our method can effectively segment sounding objects without being biased to salient objects and also achieves state-of-the-art performance in both the single-source and multi-source scenarios.
The task of spatio-temporal video grounding involves identifying the spatial and temporal regions in a video that correspond to the objects or actions described in a given textual description. However, current models used for spatio-temporal video grounding often rely heavily on spatio-temporal priors to make the predictions. As a result, they may suffer from spurious correlations and lack the ability to generalize well to new or diverse scenarios. To overcome this limitation, we introduce a deconfounded multimodal learning framework, which utilizes a structural causal model to treat dataset biases as a confounder and subsequently remove their confounding effect. Through this framework, we can perform causal intervention on the multimodal input and derive an unbiased estimation formula through the do-calculus technique. In order to tackle the challenge of diverse and often unobservable confounders, we further propose a novel retrieval-based approach with a causal mask mechanism. The proposed method leverages analogical reasoning to facilitate deconfounded learning and mitigate dataset biases, enabling unbiased spatio-temporal prediction without explicitly modeling the confounding factors. Extensive experiments on two challenging benchmarks have well verified the effectiveness and rationality of our proposed solution.
Graph anomaly detection (GAD) has attracted increasing attention in machine learning and data mining. Recent works have mainly focused on how to capture richer information to improve the quality of node embeddings for GAD. Despite their significant advances in detection performance, there is still a relative dearth of research on the properties of the task. GAD aims to discern the anomalies that deviate from most nodes. However, the model is prone to learn the pattern of normal samples which make up the majority of samples. Meanwhile, anomalies can be easily detected when their behaviors differ from normality. Therefore, the performance can be further improved by enhancing the ability to learn the normal pattern. To this end, we propose a normality learning-based GAD framework via multi-scale contrastive learning networks (NLGAD for abbreviation). Specifically, we first initialize the model with the contrastive networks on different scales. To provide sufficient and reliable normal nodes for normality learning, we design an effective hybrid strategy for normality selection. Finally, the model is refined with the only input of reliable normal nodes and learns a more accurate estimate of normality so that anomalous nodes can be more easily distinguished. Eventually, extensive experiments on six benchmark graph datasets demonstrate the effectiveness of our normality learning-based scheme on GAD. Notably, the proposed algorithm improves the detection performance (up to 5.89% AUC gain) compared with the state-of-the-art methods. The source code is released at https://github.com/FelixDJC/NLGAD.
SimulFlow: Simultaneously Extracting Feature and Identifying Target for Unsupervised Video Object Segmentation
PDF ↗Unsupervised video object segmentation (UVOS) aims at detecting the primary objects in a given video sequence without any human interposing. Most existing methods rely on two-stream architectures that separately encode the appearance and motion information before fusing them to identify the target and generate object masks. However, this pipeline is computationally expensive and can lead to suboptimal performance due to the difficulty of fusing the two modalities properly. In this paper, we propose a novel UVOS model called SimulFlow that simultaneously performs feature extraction and target identification, enabling efficient and effective unsupervised video object segmentation. Concretely, we design a novel SimulFlow Attention mechanism to bridege the image and motion by utilizing the flexibility of attention operation, where coarse masks predicted from fused feature at each stage are used to constrain the attention operation within the mask area and exclude the impact of noise. Because of the bidirectional information flow between visual and optical flow features in SimulFlow Attention, no extra hand-designed fusing module is required and we only adopt a light decoder to obtain the final prediction. We evaluate our method on several benchmark datasets and achieve state-of-the-art results. Our proposed approach not only outperforms existing methods but also addresses the computational complexity and fusion difficulties caused by two-stream architectures. Our models achieve 87.4 ℐ&F on DAVIS-16 with the highest speed (63.7 FPS on a 3090) and the lowest parameters (13.7 M). Our SimulFlow also obtains competitive results on video salient object detection datasets.
Salient Object Detection (SOD) aims to identify and segment the most conspicuous objects in an image or video. As an important pre-processing step, it has many potential applications in multimedia and vision tasks. With the advance of imaging devices, SOD with high-resolution images is of great demand, recently. However, traditional SOD methods are largely limited to low-resolution images, making them difficult to adapt to the development of High-Resolution SOD (HRSOD). Although some HRSOD methods emerge, there are no large enough datasets for training and evaluating. Besides, current HRSOD methods generally produce incomplete object regions and irregular object boundaries. To address above issues, in this work, we first propose a new HRS10K dataset, which contains 10,500 high-quality annotated images at 2K-8K resolution. As far as we know, it is the largest dataset for the HRSOD task, which will significantly help future works in training and evaluating models. Furthermore, to improve the HRSOD performance, we propose a novel Recurrent Multi-scale Transformer (RMFormer), which recurrently utilizes shared Transformers and multi-scale refinement architectures. Thus, high-resolution saliency maps can be generated with the guidance of lower-resolution predictions. Extensive experiments on both high-resolution and low-resolution benchmarks show the effectiveness and superiority of the proposed framework. The source code and dataset are released at: https://github.com/DrowsyMon/RMFormer.
A Capture to Registration Framework for Realistic Image Super-Resolution in the Industry Environment
The acquisition and processing of visual data in industrial environments are of paramount importance. High-resolution (HR) images offer superior clarity and richer textural detail compared to low-resolution (LR) images. On the one hand, owing to the incorporation of richer information, HR images demonstrate substantially enhanced performance compared to LR images in downstream applications, such as anomaly detection. On the other hand, they provide valuable insights to designers and quality inspectors who require a detailed understanding of the images. Currently, the majority of research on super-resolution focuses on natural scenes such as cities and fields, however, the development of datasets for industrial scenes is still in its infancy. To address the image distortion in building realistic LR-HR image pairs in the industry environment, we design a capture to registration framework. It consists of the standard imaging system, physical calibration of the imaging system, as well as the rigid to elastic registration of the LR-HR image pairs. Thus, we build the first realistic industrial sence super-resolution dataset (IndSR), comprises of 50 sets of calibrated images with three scale factors and five typical defects. To benchmark IndSR, we employ quantitative, qualitative, and task-oriented studies to evaluate the representative super-resolution and anomaly detection methods. Besides, we systematically investigate and discuss the performances and results of the existing SISR methods to advance research in the field of super-resolution in industry environment. The IndSR dataset can be available from https://byw4ng.github.io/IndSR/.
Federated learning (FL) is a nascent distributed learning paradigm to train a shared global model without violating users' privacy. FL has been shown to be vulnerable to various Byzantine attacks, where malicious participants could independently or collusively upload well-crafted updates to deteriorate the performance of the global model. However, existing defenses could only mitigate part of Byzantine attacks, without providing an all-sided shield for FL. It is difficult to simply combine them as they rely on totally contradictory assumptions. In this paper, we propose FPD, a four-pronged defense against both non-colluding and colluding Byzantine attacks. Our main idea is to utilize absolute similarity to filter updates rather than relative similarity used in existingI works. To this end, we first propose a reliable client selection strategy to prevent the majority of threats in the bud. Then we design a simple but effective score-based detection method to mitigate colluding attacks. Third, we construct an enhanced spectral-based outlier detector to accurately discard abnormal updates when the training data is not independent and identically distributed (non-IID). Finally, we design update denoising to rectify the direction of the slightly noisy but harmful updates. The four sequentially combined modules can effectively reconcile the contradiction in addressing non-colluding and colluding Byzantine attacks. Extensive experiments over three benchmark image classification datasets against four state-of-the-art Byzantine attacks demonstrate that FPD drastically outperforms existing defenses in IID and non-IID scenarios (with 30% improvement on model accuracy).
Retinal fundus images have been applied for the diagnosis and screening of eye diseases, such as Diabetic Retinopathy (DR) or Diabetic Macular Edema (DME). However, both low-quality fundus images and style inconsistency potentially increase uncertainty in the diagnosis of fundus disease and even lead to misdiagnosis by ophthalmologists. Most of the existing fundus image enhancement methods mainly focus on improving the image quality by leveraging the guidance of high-quality images, which is difficult to be collected in medical applications. In this paper, we tackle image quality enhancement in a fully unsupervised setting, i.e., neither paired images nor high-quality images. To this end, we explore the potential of the self-supervised task for improving the quality of fundus images without the requirement of high-quality reference images, and proposed a Domain Adaptation Self-supervised Quality Enhancement framework, named DASQE. Specifically, we construct multiple patch-wise domains via a well-designed rule-based quality assessment scheme and style clustering. To achieve robust low-quality image enhancement and address style inconsistency, we formulate two self-supervised domain adaptation tasks to disentangle the features of image content, low-quality factors and style information by exploring intrinsic supervision signals within the low-quality images. Extensive experiments are conducted on four benchmark datasets, and results show that our DASQE method achieves new state-of-the-art performance when only low-quality images are available.
Deepfake has taken the world by storm, triggering a trust crisis. Current deepfake detection methods are typically inadequate in generalizability, with a tendency to overfit to image contents such as the background, which are frequently occurring but relatively unimportant in the training dataset. Furthermore, current methods heavily rely on a few dominant forgery regions and may ignore other equally important regions, leading to inadequate uncovering of forgery cues. In this paper, we strive to address these shortcomings from three aspects: (1) We propose an innovative two-stream network that effectively enlarges the potential regions from which the model extracts forgery evidence. (2) We devise three functional modules to handle the multi-stream and multi-scale features in a collaborative learning scheme. (3) Confronted with the challenge of obtaining forgery annotations, we propose a Semi-supervised Patch Similarity Learning strategy to estimate patch-level forged location annotations. Empirically, our method demonstrates significantly improved robustness and generalizability, outperforming previous methods on six benchmarks, and improving the frame-level AUC on Deepfake Detection Challenge preview dataset from 0.797 to 0.835 and video-level AUC on CelebDF_v1 dataset from 0.811 to 0.847. Our implementation is available at https://github.com/sccsok/Locate-and-Verify.
Exploring Shape Embedding for Cloth-Changing Person Re-Identification via 2D-3D Correspondences
PDF ↗Cloth-Changing Person Re-Identification (CC-ReID) is a common and realistic problem since fashion constantly changes over time and people's aesthetic preferences are not set in stone. While most existing cloth-changing ReID methods focus on learning cloth-agnostic identity representations from coarse semantic cues (e.g. silhouettes and part segmentation maps), they neglect the continuous shape distributions at the pixel level. In this paper, we propose Continuous Surface Correspondence Learning (CSCL), a new shape embedding paradigm for cloth-changing ReID. CSCL establishes continuous correspondences between a 2D image plane and a canonical 3D body surface via pixel-to-vertex classification, which naturally aligns a person image to the surface of a 3D human model and simultaneously obtains pixel-wise surface embeddings. We further extract fine-grained shape features from the learned surface embeddings and then integrate them with global RGB features via a carefully designed cross-modality fusion module. The shape embedding paradigm based on 2D-3D correspondences remarkably enhances the model's global understanding of human body shape. To promote the study of ReID under clothing change, we construct 3D Dense Persons (DP3D), which is the first large-scale cloth-changing ReID dataset that provides densely annotated 2D-3D correspondences and a precise 3D mesh for each person image, while containing diverse cloth-changing cases over all four seasons. Experiments on both cloth-changing and cloth-consistent ReID benchmarks validate the effectiveness of our method.
Personalized fashion matching task aims to predict the compatible fashion items given available ones for specific users through the effective modeling of the third-order interaction patterns among the user and item pairs. To achieve this, previous methods separately model two key components, user-item and item-item relationships, which ignore the inherent correlations between them and lead to undesirable performance. With a new perspective, this paper proposes to formulate the personalized item matching as the multi-relational connectivity and apply a single-component translation operation to model the targeted third-order interactions. With user-item-item interactions naturally constructing a multi-relational graph, we further device two graph learning modules to enhance the translation-based matching approach from two perspectives,C ontext and Path. The proposed method, named CP-TransMatch, has been tested with extensive experiments on three benchmark fashion datasets and proven effective. It sets the new SOTA for the personalized fashion matching task.
Multi-modal hashing (MMH) encodes multi-modal data into latent hash code, and has been widely applied for efficient large-scale multi-modal retrieval. In practice it is common that multi-modal data is often corrupted with missing modalities, e.g., social image often lacks its tags in image-text retrieval. Conventional MMHs can only learn on complete modalities, which however wastes a considerable amount of collected data. To fulfill this gap, this paper proposes Graph Convolutional Incomplete Multi-modal Hashing (GCIMH) to learn hash code on incomplete multi-modal data. GCIMH develops Graph Convolutional Autoencoder to reconstruct incomplete multi-modal data with effective exploit of its semantic structure. GCIMH further develops multi-modal and label networks to encode multiple modalities and label respectively. GCIMH can successfully transfer knowledge of autoencoder and label network to multi-modal hashing network using teacher-student learning framework. GCIMH can handle missing modalities in both offline training and online query stages. Extensive empirical studies on three benchmark datasets demonstrate the superiority of the proposed GCIMH over the state-of-the-arts on both complete and incomplete multi-modal retrieval.
Toward Zero-shot Character Recognition: A Gold Standard Dataset with Radical-level Annotations
PDF ↗Optical character recognition (OCR) methods have been applied to diverse tasks, e.g., street view text recognition and document analysis. Recently, zero-shot OCR has piqued the interest of the research community because it considers a practical OCR scenario with unbalanced data distribution. However, there is a lack of benchmarks for evaluating such zero-shot methods that apply a divide-and-conquer recognition strategy by decomposing characters into radicals. Meanwhile, radical recognition, as another important OCR task, also lacks radical-level annotation for model training. In this paper, we construct an ancient Chinese character image dataset that contains both radical-level and character-level annotations to satisfy the requirements of the above-mentioned methods, namely, ACCID, where radical-level annotations include radical categories, radical locations, and structural relations. To increase the adaptability of ACCID, we propose a splicing-based synthetic character algorithm to augment the training samples and apply an image denoising method to improve the image quality. By introducing character decomposition and recombination, we propose a baseline method for zero-shot OCR. The experimental results demonstrate the validity of ACCID and the baseline model quantitatively and qualitatively.
Composite images are generated by combining two or more different photographs, and their content is typically heterogeneous. However, existing unimodal visual redundancy prediction methods are difficult to accurately model the complex characteristics of this image type. In this paper, we investigate the visual redundancy modeling of composite images from an end-to-end multimodal perspective, including four cross-media modalities (i.e., text, brightness, color, and segmentation). Specifically, we design a two-stage cross-modal alignment module based on self-attention mechanism and contrastive learning, and develop a fusion module based on a cross-modal augmentation paradigm. Further, we establish the first cross-media visual redundancy dataset for composite images, which contains 413 groups of cross-modal data and generates 13629 realistic compression distortions using the latest versatile video coding (VVC) standard. Experimental results on nine benchmark datasets demonstrate the effectiveness of our method, outperforming seven representative methods.
Video quality assessment (VQA) has attracted growing attention in recent years. While the great expense of annotating large-scale VQA datasets has become the main obstacle for current deep-learning methods. To surmount the constraint of insufficient training data, in this paper, we first consider the complete range of video distribution diversity (i.e. content, distortion, motion) and employ diverse pretrained models (e.g. architecture, pretext task, pre-training dataset) to benefit quality representation. An Adaptive Diverse Quality-aware feature Acquisition (Ada-DQA) framework is proposed to capture desired quality-related features generated by these frozen pretrained models. By leveraging the Quality-aware Acquisition Module (QAM), the framework is able to extract more essential and relevant features to represent quality. Finally, the learned quality representation is utilized as supplementary supervisory information, along with the supervision of the labeled quality score, to guide the training of a relatively lightweight VQA model in a knowledge distillation manner, which largely reduces the computational cost during inference. Experimental results on three mainstream no-reference VQA benchmarks clearly show the superior performance of Ada-DQA in comparison with current state-of-the-art approaches without using extra training data of VQA.