Authors: Rasha Alshawi, Md Meftahul Ferdaus, Md Tamjidul Hoque, Kendall Niles, Ken Pathak, Steve Sloan, Mahdi Abdelguerfi
Abstract: This paper introduces Semantic Haar-Adaptive Refined Pyramid Network (SHARP-Net), a novel architecture for semantic segmentation. SHARP-Net integrates a bottom-up pathway featuring Inception-like blocks with varying filter sizes (3x3$ and 5x5), parallel max-pooling, and additional spatial detection layers. This design captures multi-scale features and fine structural details. Throughout the network, depth-wise separable convolutions are used to reduce complexity. The top-down pathway of SHARP-Net focuses on generating high-resolution features through upsampling and information fusion using $1\times1$ and $3\times3$ depth-wise separable convolutions. We evaluated our model using our developed challenging Culvert-Sewer Defects dataset and the benchmark DeepGlobe Land Cover dataset. Our experimental evaluation demonstrated the base model's (excluding Haar-like features) effectiveness in handling irregular defect shapes, occlusions, and class imbalances. It outperformed state-of-the-art methods, including U-Net, CBAM U-Net, ASCU-Net, FPN, and SegFormer, achieving average improvements of 14.4% and 12.1% on the Culvert-Sewer Defects and DeepGlobe Land Cover datasets, respectively, with IoU scores of 77.2% and 70.6%. Additionally, the training time was reduced. Furthermore, the integration of carefully selected and fine-tuned Haar-like features enhanced the performance of deep learning models by at least 20%. The proposed SHARP-Net, incorporating Haar-like features, achieved an impressive IoU of 94.75%, representing a 22.74% improvement over the base model. These features were also applied to other deep learning models, showing a 35.0% improvement, proving their versatility and effectiveness. SHARP-Net thus provides a powerful and efficient solution for accurate semantic segmentation in challenging real-world scenarios.
Authors: Mikyla K. Bowen, Jesse W. Wilson
Abstract: We propose a novel application of image translation networks for image classification and demonstrate its potential as a more interpretable alternative to conventional black box classification networks. We train a network to translate images between possible classes, and then quantify translation distance, i.e. the degree of alteration needed to conform an image to one class or another. These translation distances can then be examined for clusters and trends, and can be fed directly to a simple classifier (e.g. a support vector machine, SVM), providing comparable accuracy compared to a conventional end-to-end convolutional neural network classifier. In addition, visual inspection of translated images can reveal class-specific characteristics and biases in the training sets, such as visual artifacts that are more frequently observed in one class or another. We demonstrate the approach on a toy 2-class scenario, apples versus oranges, and then apply it to two medical imaging tasks: detecting melanoma from photographs of pigmented lesions and classifying 6 cell types in a bone marrow biopsy smear. This novel application of image-to-image networks shows the potential of the technology to go beyond imagining different stylistic changes and to provide greater insight into image classification and medical imaging datasets.
Authors: Vinit Hegiste, Snehal Walunj, Jibinraj Antony, Tatjana Legler, Martin Ruskowski
Abstract: Federated Learning (FL) has garnered significant attention in manufacturing for its robust model development and privacy-preserving capabilities. This paper contributes to research focused on the robustness of FL models in object detection, hereby presenting a comparative study with conventional techniques using a hybrid dataset for small object detection. Our findings demonstrate the superior performance of FL over centralized training models and different deep learning techniques when tested on test data recorded in a different environment with a variety of object viewpoints, lighting conditions, cluttered backgrounds, etc. These results highlight the potential of FL in achieving robust global models that perform efficiently even in unseen environments. The study provides valuable insights for deploying resilient object detection models in manufacturing environments.
Authors: Simon Deltadahl, Julian Gilbey, Christine Van Laer, Nancy Boeckx, Mathie Leers, Tanya Freeman, Laura Aiken, Timothy Farren, Matthew Smith, Mohamad Zeina, BloodCounts! consortium, Concetta Piazzese, Joseph Taylor, Nicholas Gleadall, Carola-Bibiane Sch\"onlieb, Suthesh Sivapalaratnam, Michael Roberts, Parashkev Nachev
Abstract: Accurate classification of haematological cells is critical for diagnosing blood disorders, but presents significant challenges for machine automation owing to the complexity of cell morphology, heterogeneities of biological, pathological, and imaging characteristics, and the imbalance of cell type frequencies. We introduce CytoDiffusion, a diffusion-based classifier that effectively models blood cell morphology, combining accurate classification with robust anomaly detection, resistance to distributional shifts, interpretability, data efficiency, and superhuman uncertainty quantification. Our approach outperforms state-of-the-art discriminative models in anomaly detection (AUC 0.976 vs. 0.919), resistance to domain shifts (85.85% vs. 74.38% balanced accuracy), and performance in low-data regimes (95.88% vs. 94.95% balanced accuracy). Notably, our model generates synthetic blood cell images that are nearly indistinguishable from real images, as demonstrated by a Turing test in which expert haematologists achieved only 52.3% accuracy (95% CI: [50.5%, 54.2%]). Furthermore, we enhance model explainability through the generation of directly interpretable counterfactual heatmaps. Our comprehensive evaluation framework, encompassing these multiple performance dimensions, establishes a new benchmark for medical image analysis in haematology, ultimately enabling improved diagnostic accuracy in clinical settings. Our code is available at https://github.com/Deltadahl/CytoDiffusion.
Authors: Daryn Sagel, Bryan Quaife
Abstract: The increasing frequency and severity of wildfires highlight the need for accurate fire and plume spread models. We introduce an approach that effectively isolates and tracks fire and plume behavior across various spatial and temporal scales and image types, identifying physical phenomena in the system and providing insights useful for developing and validating models. Our method combines image segmentation and graph theory to delineate fire fronts and plume boundaries. We demonstrate that the method effectively distinguishes fires and plumes from visually similar objects. Results demonstrate the successful isolation and tracking of fire and plume dynamics across various image sources, ranging from synoptic-scale ($10^4$-$10^5$ m) satellite images to sub-microscale ($10^0$-$10^1$ m) images captured close to the fire environment. Furthermore, the methodology leverages image inpainting and spatio-temporal dataset generation for use in statistical and machine learning models.
Authors: Nayeem Ahmed, Md Maruf Rahman, Md Fatin Ishrak, Md Imran Kabir Joy, Md Sanowar Hossain Sabuj, Md. Sadekur Rahman
Abstract: This study compares eight pre-trained CNNs for diagnosing keratoconus, a degenerative eye disease. A carefully selected dataset of keratoconus, normal, and suspicious cases was used. The models tested include DenseNet121, EfficientNetB0, InceptionResNetV2, InceptionV3, MobileNetV2, ResNet50, VGG16, and VGG19. To maximize model training, bad sample removal, resizing, rescaling, and augmentation were used. The models were trained with similar parameters, activation function, classification function, and optimizer to compare performance. To determine class separation effectiveness, each model was evaluated on accuracy, precision, recall, and F1-score. MobileNetV2 was the best accurate model in identifying keratoconus and normal cases with few misclassifications. InceptionV3 and DenseNet121 both performed well in keratoconus detection, but they had trouble with questionable cases. In contrast, EfficientNetB0, ResNet50, and VGG19 had more difficulty distinguishing dubious cases from regular ones, indicating the need for model refining and development. A detailed comparison of state-of-the-art CNN architectures for automated keratoconus identification reveals each model's benefits and weaknesses. This study shows that advanced deep learning models can enhance keratoconus diagnosis and treatment planning. Future research should explore hybrid models and integrate clinical parameters to improve diagnostic accuracy and robustness in real-world clinical applications, paving the way for more effective AI-driven ophthalmology tools.
Authors: Muhammad Haseeb Aslam, Marco Pedersoli, Alessandro Lameiras Koerich, Eric Granger
Abstract: Human emotion is a complex phenomenon conveyed and perceived through facial expressions, vocal tones, body language, and physiological signals. Multimodal emotion recognition systems can perform well because they can learn complementary and redundant semantic information from diverse sensors. In real-world scenarios, only a subset of the modalities employed for training may be available at test time. Learning privileged information allows a model to exploit data from additional modalities that are only available during training. SOTA methods for PKD have been proposed to distill information from a teacher model (with privileged modalities) to a student model (without privileged modalities). However, such PKD methods utilize point-to-point matching and do not explicitly capture the relational information. Recently, methods have been proposed to distill the structural information. However, PKD methods based on structural similarity are primarily confined to learning from a single joint teacher representation, which limits their robustness, accuracy, and ability to learn from diverse multimodal sources. In this paper, a multi-teacher PKD (MT-PKDOT) method with self-distillation is introduced to align diverse teacher representations before distilling them to the student. MT-PKDOT employs a structural similarity KD mechanism based on a regularized optimal transport (OT) for distillation. The proposed MT-PKDOT method was validated on the Affwild2 and Biovid datasets. Results indicate that our proposed method can outperform SOTA PKD methods. It improves the visual-only baseline on Biovid data by 5.5%. On the Affwild2 dataset, the proposed method improves 3% and 5% over the visual-only baseline for valence and arousal respectively. Allowing the student to learn from multiple diverse sources is shown to increase the accuracy and implicitly avoids negative transfer to the student model.
Authors: Hao Tang, Weiyao Wang, Pierre Gleize, Matt Feiszli
Abstract: Recovering camera poses from a set of images is a foundational task in 3D computer vision, which powers key applications such as 3D scene/object reconstructions. Classic methods often depend on feature correspondence, such as keypoints, which require the input images to have large overlap and small viewpoint changes. Such requirements present considerable challenges in scenarios with sparse views. Recent data-driven approaches aim to directly output camera poses, either through regressing the 6DoF camera poses or formulating rotation as a probability distribution. However, each approach has its limitations. On one hand, directly regressing the camera poses can be ill-posed, since it assumes a single mode, which is not true under symmetry and leads to sub-optimal solutions. On the other hand, probabilistic approaches are capable of modeling the symmetry ambiguity, yet they sample the entire space of rotation uniformly by brute-force. This leads to an inevitable trade-off between high sample density, which improves model precision, and sample efficiency that determines the runtime. In this paper, we propose ADen to unify the two frameworks by employing a generator and a discriminator: the generator is trained to output multiple hypotheses of 6DoF camera pose to represent a distribution and handle multi-mode ambiguity, and the discriminator is trained to identify the hypothesis that best explains the data. This allows ADen to combine the best of both worlds, achieving substantially higher precision as well as lower runtime than previous methods in empirical evaluations.
Authors: Zhiyi Shi, Junsik Kim, Wanhua Li, Yicong Li, Hanspeter Pfister
Abstract: Multi-modal pre-trained models efficiently extract and fuse features from different modalities with low memory requirements for fine-tuning. Despite this efficiency, their application in disease diagnosis is under-explored. A significant challenge is the frequent occurrence of missing modalities, which impairs performance. Additionally, fine-tuning the entire pre-trained model demands substantial computational resources. To address these issues, we introduce Modality-aware Low-Rank Adaptation (MoRA), a computationally efficient method. MoRA projects each input to a low intrinsic dimension but uses different modality-aware up-projections for modality-specific adaptation in cases of missing modalities. Practically, MoRA integrates into the first block of the model, significantly improving performance when a modality is missing. It requires minimal computational resources, with less than 1.6% of the trainable parameters needed compared to training the entire model. Experimental results show that MoRA outperforms existing techniques in disease diagnosis, demonstrating superior performance, robustness, and training efficiency.
Authors: Shashank Kotyan, Pin-Yu Chen, Danilo Vasconcellos Vargas
Abstract: Most evaluations of vision models use indirect methods to assess latent space quality. These methods often involve adding extra layers to project the latent space into a new one. This projection makes it difficult to analyze and compare the original latent space. This article uses the k* Distribution, a local neighborhood analysis method, to examine the learned latent space at the level of individual concepts, which can be extended to examine the entire latent space. We introduce skewness-based true and approximate metrics for interpreting individual concepts to assess the overall quality of vision models' latent space. Our findings indicate that current vision models frequently fracture the distributions of individual concepts within the latent space. Nevertheless, as these models improve in generalization across multiple datasets, the degree of fracturing diminishes. A similar trend is observed in robust vision models, where increased robustness correlates with reduced fracturing. Ultimately, this approach enables a direct interpretation and comparison of the latent spaces of different vision models and reveals a relationship between a model's generalizability and robustness. Results show that as a model becomes more general and robust, it tends to learn features that result in better clustering of concepts. Project Website is available online at https://shashankkotyan.github.io/k-Distribution/
Authors: Aoran Xiao, Weihao Xuan, Heli Qi, Yun Xing, Naoto Yokoya, Shijian Lu
Abstract: Robust and accurate segmentation of scenes has become one core functionality in various visual recognition and navigation tasks. This has inspired the recent development of Segment Anything Model (SAM), a foundation model for general mask segmentation. However, SAM is largely tailored for single-modal RGB images, limiting its applicability to multi-modal data captured with widely-adopted sensor suites, such as LiDAR plus RGB, depth plus RGB, thermal plus RGB, etc. We develop MM-SAM, an extension and expansion of SAM that supports cross-modal and multi-modal processing for robust and enhanced segmentation with different sensor suites. MM-SAM features two key designs, namely, unsupervised cross-modal transfer and weakly-supervised multi-modal fusion, enabling label-efficient and parameter-efficient adaptation toward various sensor modalities. It addresses three main challenges: 1) adaptation toward diverse non-RGB sensors for single-modal processing, 2) synergistic processing of multi-modal data via sensor fusion, and 3) mask-free training for different downstream tasks. Extensive experiments show that MM-SAM consistently outperforms SAM by large margins, demonstrating its effectiveness and robustness across various sensors and data modalities.
Authors: Wei Sun, Yuan Li, Qixiang Ye, Jianbin Jiao, Yanzhao Zhou
Abstract: Depth information provides valuable insights into the 3D structure especially the outline of objects, which can be utilized to improve the semantic segmentation tasks. However, a naive fusion of depth information can disrupt feature and compromise accuracy due to the modality gap between the depth and the vision. In this work, we introduce a Depth-guided Texture Diffusion approach that effectively tackles the outlined challenge. Our method extracts low-level features from edges and textures to create a texture image. This image is then selectively diffused across the depth map, enhancing structural information vital for precisely extracting object outlines. By integrating this enriched depth map with the original RGB image into a joint feature embedding, our method effectively bridges the disparity between the depth map and the image, enabling more accurate semantic segmentation. We conduct comprehensive experiments across diverse, commonly-used datasets spanning a wide range of semantic segmentation tasks, including Camouflaged Object Detection (COD), Salient Object Detection (SOD), and indoor semantic segmentation. With source-free estimated depth or depth captured by depth cameras, our method consistently outperforms existing baselines and achieves new state-of-theart results, demonstrating the effectiveness of our Depth-guided Texture Diffusion for image semantic segmentation.
Authors: Xiao Zhao, Bo Chen, Mingyang Sun, Dingkang Yang, Youxing Wang, Xukun Zhang, Mingcheng Li, Dongliang Kou, Xiaoyi Wei, Lihua Zhang
Abstract: Vision-based 3D semantic scene completion (SSC) describes autonomous driving scenes through 3D volume representations. However, the occlusion of invisible voxels by scene surfaces poses challenges to current SSC methods in hallucinating refined 3D geometry. This paper proposes HybridOcc, a hybrid 3D volume query proposal method generated by Transformer framework and NeRF representation and refined in a coarse-to-fine SSC prediction framework. HybridOcc aggregates contextual features through the Transformer paradigm based on hybrid query proposals while combining it with NeRF representation to obtain depth supervision. The Transformer branch contains multiple scales and uses spatial cross-attention for 2D to 3D transformation. The newly designed NeRF branch implicitly infers scene occupancy through volume rendering, including visible and invisible voxels, and explicitly captures scene depth rather than generating RGB color. Furthermore, we present an innovative occupancy-aware ray sampling method to orient the SSC task instead of focusing on the scene surface, further improving the overall performance. Extensive experiments on nuScenes and SemanticKITTI datasets demonstrate the effectiveness of our HybridOcc on the SSC task.
Authors: Jiancheng Pan, Yanxing Liu, Yuqian Fu, Muyuan Ma, Jiaohao Li, Danda Pani Paudel, Luc Van Gool, Xiaomeng Huang
Abstract: Object detection, particularly open-vocabulary object detection, plays a crucial role in Earth sciences, such as environmental monitoring, natural disaster assessment, and land-use planning. However, existing open-vocabulary detectors, primarily trained on natural-world images, struggle to generalize to remote sensing images due to a significant data domain gap. Thus, this paper aims to advance the development of open-vocabulary object detection in remote sensing community. To achieve this, we first reformulate the task as Locate Anything on Earth (LAE) with the goal of detecting any novel concepts on Earth. We then developed the LAE-Label Engine which collects, auto-annotates, and unifies up to 10 remote sensing datasets creating the LAE-1M - the first large-scale remote sensing object detection dataset with broad category coverage. Using the LAE-1M, we further propose and train the novel LAE-DINO Model, the first open-vocabulary foundation object detector for the LAE task, featuring Dynamic Vocabulary Construction (DVC) and Visual-Guided Text Prompt Learning (VisGT) modules. DVC dynamically constructs vocabulary for each training batch, while VisGT maps visual features to semantic space, enhancing text features. We comprehensively conduct experiments on established remote sensing benchmark DIOR, DOTAv2.0, as well as our newly introduced 80-class LAE-80C benchmark. Results demonstrate the advantages of the LAE-1M dataset and the effectiveness of the LAE-DINO method.
Authors: Weiming Zhang, Yexin Liu, Xu Zheng, Lin Wang
Abstract: This paper presents GoodSAM++, a novel framework utilizing the powerful zero-shot instance segmentation capability of SAM (i.e., teacher) to learn a compact panoramic semantic segmentation model, i.e., student, without requiring any labeled data. GoodSAM++ addresses two critical challenges: 1) SAM's inability to provide semantic labels and inherent distortion problems of panoramic images; 2) the significant capacity disparity between SAM and the student. The `out-of-the-box' insight of GoodSAM++ is to introduce a teacher assistant (TA) to provide semantic information for SAM, integrated with SAM to obtain reliable pseudo semantic maps to bridge both domain and capacity gaps. To make this possible, we first propose a Distortion-Aware Rectification (DARv2) module to address the domain gap. It effectively mitigates the object deformation and distortion problem in panoramic images to obtain pseudo semantic maps. We then introduce a Multi-level Knowledge Adaptation (MKA) module to efficiently transfer the semantic information from the TA and pseudo semantic maps to our compact student model, addressing the significant capacity gap. We conduct extensive experiments on both outdoor and indoor benchmark datasets, showing that our GoodSAM++ achieves a remarkable performance improvement over the state-of-the-art (SOTA) domain adaptation methods. Moreover, diverse open-world scenarios demonstrate the generalization capacity of our GoodSAM++. Last but not least, our most lightweight student model achieves comparable performance to the SOTA models with only 3.7 million parameters.
Authors: Aayush Agrawal, Ashmitha Jaysi Sivakumar, Ibrahim Kaif, Chayan Banerjee
Abstract: Accurate lane detection is essential for effective path planning and lane following in autonomous driving, especially in scenarios with significant occlusion from vehicles and pedestrians. Existing models often struggle under such conditions, leading to unreliable navigation and safety risks. We propose two innovative approaches to enhance lane detection in these challenging environments, each showing notable improvements over current methods. The first approach aug-Segment improves conventional lane detection models by augmenting the training dataset of CULanes with simulated occlusions and training a segmentation model. This method achieves a 12% improvement over a number of SOTA models on the CULanes dataset, demonstrating that enriched training data can better handle occlusions, however, since this model lacked robustness to certain settings, our main contribution is the second approach, LOID Lane Occlusion Inpainting and Detection. LOID introduces an advanced lane detection network that uses an image processing pipeline to identify and mask occlusions. It then employs inpainting models to reconstruct the road environment in the occluded areas. The enhanced image is processed by a lane detection algorithm, resulting in a 20% & 24% improvement over several SOTA models on the BDDK100 and CULanes datasets respectively, highlighting the effectiveness of this novel technique.
Authors: Xiao Zhao, Xukun Zhang, Dingkang Yang, Mingyang Sun, Mingcheng Li, Shunli Wang, Lihua Zhang
Abstract: Accurate and robust multimodal multi-task perception is crucial for modern autonomous driving systems. However, current multimodal perception research follows independent paradigms designed for specific perception tasks, leading to a lack of complementary learning among tasks and decreased performance in multi-task learning (MTL) due to joint training. In this paper, we propose MaskBEV, a masked attention-based MTL paradigm that unifies 3D object detection and bird's eye view (BEV) map segmentation. MaskBEV introduces a task-agnostic Transformer decoder to process these diverse tasks, enabling MTL to be completed in a unified decoder without requiring additional design of specific task heads. To fully exploit the complementary information between BEV map segmentation and 3D object detection tasks in BEV space, we propose spatial modulation and scene-level context aggregation strategies. These strategies consider the inherent dependencies between BEV segmentation and 3D detection, naturally boosting MTL performance. Extensive experiments on nuScenes dataset show that compared with previous state-of-the-art MTL methods, MaskBEV achieves 1.3 NDS improvement in 3D object detection and 2.7 mIoU improvement in BEV map segmentation, while also demonstrating slightly leading inference speed.
Authors: Xiaokun Sun, Zhenyu Zhang, Ying Tai, Qian Wang, Hao Tang, Zili Yi, Jian Yang
Abstract: Recent advances in text-guided 3D avatar generation have made substantial progress by distilling knowledge from diffusion models. Despite the plausible generated appearance, existing methods cannot achieve fine-grained disentanglement or high-fidelity modeling between inner body and outfit. In this paper, we propose Barbie, a novel framework for generating 3D avatars that can be dressed in diverse and high-quality Barbie-like garments and accessories. Instead of relying on a holistic model, Barbie achieves fine-grained disentanglement on avatars by semantic-aligned separated models for human body and outfits. These disentangled 3D representations are then optimized by different expert models to guarantee the domain-specific fidelity. To balance geometry diversity and reasonableness, we propose a series of losses for template-preserving and human-prior evolving. The final avatar is enhanced by unified texture refinement for superior texture consistency. Extensive experiments demonstrate that Barbie outperforms existing methods in both dressed human and outfit generation, supporting flexible apparel combination and animation. The code will be released for research purposes. Our project page is: https://2017211801.github.io/barbie.github.io/.
Authors: Sheng Ye, Zhen-Hui Dong, Yubin Hu, Yu-Hui Wen, Yong-Jin Liu
Abstract: 3D Gaussian Splatting has recently emerged as a powerful representation that can synthesize remarkable novel views using consistent multi-view images as input. However, we notice that images captured in dark environments where the scenes are not fully illuminated can exhibit considerable brightness variations and multi-view inconsistency, which poses great challenges to 3D Gaussian Splatting and severely degrades its performance. To tackle this problem, we propose Gaussian-DK. Observing that inconsistencies are mainly caused by camera imaging, we represent a consistent radiance field of the physical world using a set of anisotropic 3D Gaussians, and design a camera response module to compensate for multi-view inconsistencies. We also introduce a step-based gradient scaling strategy to constrain Gaussians near the camera, which turn out to be floaters, from splitting and cloning. Experiments on our proposed benchmark dataset demonstrate that Gaussian-DK produces high-quality renderings without ghosting and floater artifacts and significantly outperforms existing methods. Furthermore, we can also synthesize light-up images \dzh{by controlling exposure levels that clearly show details in shadow areas.
Authors: Tianyi Zhu, Wei Shang, Dongwei Ren, Wangmeng Zuo
Abstract: 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. The code is available at \url{https://github.com/Tian-one/tps-inbetween}.
Authors: Jiao Zhang, Jian Xu, Xu-Yao Zhang, Cheng-Lin Liu
Abstract: In real-world applications, the sample distribution at the inference stage often differs from the one at the training stage, causing performance degradation of trained deep models. The research on domain generalization (DG) aims to develop robust algorithms that can improve the generalized performance in unseen domains by training on a few domains. However, the domain-agnostic vision model, trained on a limited number of domains using traditional domain generalization methods, cannot guarantee its effectiveness in dealing with unseen domains. The introduction of language can break the closed cognition space of the vision model, providing additional semantic information that cannot be inferred from vision-only datasets. In this paper, we propose to overcome the challenge in previous DG methods by introducing the style prompt in the language modality to adapt the trained model dynamically. In particular, we train a style prompter to extract style information of the current image into an embedding in the token embedding space and place it in front of the candidate category words as prior knowledge to prompt the model. Our open space partition of the style token embedding space and the hand-crafted style regularization enable the trained style prompter to handle data from unknown domains effectively. Extensive experiments verify the effectiveness of our method and demonstrate state-of-the-art performances on multiple public datasets. Codes will be available after the acceptance of this paper.
Authors: Xiao Cao, Beibei Lin, Bo Wang, Zhiyong Huang, Robby T. Tan
Abstract: Sparse view NeRF is challenging because limited input images lead to an under constrained optimization problem for volume rendering. Existing methods address this issue by relying on supplementary information, such as depth maps. However, generating this supplementary information accurately remains problematic and often leads to NeRF producing images with undesired artifacts. To address these artifacts and enhance robustness, we propose SSNeRF, a sparse view semi supervised NeRF method based on a teacher student framework. Our key idea is to challenge the NeRF module with progressively severe sparse view degradation while providing high confidence pseudo labels. This approach helps the NeRF model become aware of noise and incomplete information associated with sparse views, thus improving its robustness. The novelty of SSNeRF lies in its sparse view specific augmentations and semi supervised learning mechanism. In this approach, the teacher NeRF generates novel views along with confidence scores, while the student NeRF, perturbed by the augmented input, learns from the high confidence pseudo labels. Our sparse view degradation augmentation progressively injects noise into volume rendering weights, perturbs feature maps in vulnerable layers, and simulates sparse view blurriness. These augmentation strategies force the student NeRF to recognize degradation and produce clearer rendered views. By transferring the student's parameters to the teacher, the teacher gains increased robustness in subsequent training iterations. Extensive experiments demonstrate the effectiveness of our SSNeRF in generating novel views with less sparse view degradation. We will release code upon acceptance.
Authors: Ce Wang, Wanjie Sun, Zhenzhong Chen
Abstract: Image rescaling aims to learn the optimal downscaled low-resolution (LR) image that can be accurately reconstructed to its original high-resolution (HR) counterpart. This process is crucial for efficient image processing and storage, especially in the era of ultra-high definition media. However, extreme downscaling factors pose significant challenges due to the highly ill-posed nature of the inverse upscaling process, causing existing methods to struggle in generating semantically plausible structures and perceptually rich textures. In this work, we propose a novel framework called Latent Space Based Image Rescaling (LSBIR) for extreme image rescaling tasks. LSBIR effectively leverages powerful natural image priors learned by a pre-trained text-to-image diffusion model to generate realistic HR images. The rescaling is performed in the latent space of a pre-trained image encoder and decoder, which offers better perceptual reconstruction quality due to its stronger sparsity and richer semantics. LSBIR adopts a two-stage training strategy. In the first stage, a pseudo-invertible encoder-decoder models the bidirectional mapping between the latent features of the HR image and the target-sized LR image. In the second stage, the reconstructed features from the first stage are refined by a pre-trained diffusion model to generate more faithful and visually pleasing details. Extensive experiments demonstrate the superiority of LSBIR over previous methods in both quantitative and qualitative evaluations. The code will be available at: https://github.com/wwangcece/LSBIR.
Authors: Dario Cioni, Christos Tzelepis, Lorenzo Seidenari, Ioannis Patras
Abstract: The steady improvement of Diffusion Models for visual synthesis has given rise to many new and interesting use cases of synthetic images but also has raised concerns about their potential abuse, which poses significant societal threats. To address this, fake images need to be detected and attributed to their source model, and given the frequent release of new generators, realistic applications need to consider an Open-Set scenario where some models are unseen at training time. Existing forensic techniques are either limited to Closed-Set settings or to GAN-generated images, relying on fragile frequency-based "fingerprint" features. By contrast, we propose a simple yet effective framework that incorporates features from large pre-trained foundation models to perform Open-Set origin attribution of synthetic images produced by various generative models, including Diffusion Models. We show that our method leads to remarkable attribution performance, even in the low-data regime, exceeding the performance of existing methods and generalizes better on images obtained from a diverse set of architectures. We make the code publicly available at: https://github.com/ciodar/UniversalAttribution.
Authors: Archisman Chakraborti, Bidyut B Chaudhuri
Abstract: This study introduces a novel activation function, characterized by a dynamic slope that adjusts throughout the training process, aimed at enhancing adaptability and performance in deep neural networks for computer vision tasks. The rationale behind this approach is to overcome limitations associated with traditional activation functions, such as ReLU, by providing a more flexible mechanism that can adapt to different stages of the learning process. Evaluated on the Mini-ImageNet, CIFAR-100, and MIT-BIH datasets, our method demonstrated improvements in classification metrics and generalization capabilities. These results suggest that our dynamic slope activation function could offer a new tool for improving the performance of deep learning models in various image recognition tasks.
Authors: Aniket Didolkar, Andrii Zadaianchuk, Anirudh Goyal, Mike Mozer, Yoshua Bengio, Georg Martius, Maximilian Seitzer
Abstract: The goal of object-centric representation learning is to decompose visual scenes into a structured representation that isolates the entities. Recent successes have shown that object-centric representation learning can be scaled to real-world scenes by utilizing pre-trained self-supervised features. However, so far, object-centric methods have mostly been applied in-distribution, with models trained and evaluated on the same dataset. This is in contrast to the wider trend in machine learning towards general-purpose models directly applicable to unseen data and tasks. Thus, in this work, we study current object-centric methods through the lens of zero-shot generalization by introducing a benchmark comprising eight different synthetic and real-world datasets. We analyze the factors influencing zero-shot performance and find that training on diverse real-world images improves transferability to unseen scenarios. Furthermore, inspired by the success of task-specific fine-tuning in foundation models, we introduce a novel fine-tuning strategy to adapt pre-trained vision encoders for the task of object discovery. We find that the proposed approach results in state-of-the-art performance for unsupervised object discovery, exhibiting strong zero-shot transfer to unseen datasets.
Authors: Changcheng Xiao, Qiong Cao, Zhigang Luo, Long Lan
Abstract: Tracking by detection has been the prevailing paradigm in the field of Multi-object Tracking (MOT). These methods typically rely on the Kalman Filter to estimate the future locations of objects, assuming linear object motion. However, they fall short when tracking objects exhibiting nonlinear and diverse motion in scenarios like dancing and sports. In addition, there has been limited focus on utilizing learning-based motion predictors in MOT. To address these challenges, we resort to exploring data-driven motion prediction methods. Inspired by the great expectation of state space models (SSMs), such as Mamba, in long-term sequence modeling with near-linear complexity, we introduce a Mamba-based motion model named Mamba moTion Predictor (MTP). MTP is designed to model the complex motion patterns of objects like dancers and athletes. Specifically, MTP takes the spatial-temporal location dynamics of objects as input, captures the motion pattern using a bi-Mamba encoding layer, and predicts the next motion. In real-world scenarios, objects may be missed due to occlusion or motion blur, leading to premature termination of their trajectories. To tackle this challenge, we further expand the application of MTP. We employ it in an autoregressive way to compensate for missing observations by utilizing its own predictions as inputs, thereby contributing to more consistent trajectories. Our proposed tracker, MambaTrack, demonstrates advanced performance on benchmarks such as Dancetrack and SportsMOT, which are characterized by complex motion and severe occlusion.
Authors: Jiawei Lian, Jianhong Pan, Lefan Wang, Yi Wang, Lap-Pui Chau, Shaohui Mei
Abstract: Physical attacks against object detection have gained increasing attention due to their significant practical implications. However, conducting physical experiments is extremely time-consuming and labor-intensive. Moreover, physical dynamics and cross-domain transformation are challenging to strictly regulate in the real world, leading to unaligned evaluation and comparison, severely hindering the development of physically robust models. To accommodate these challenges, we explore utilizing realistic simulation to thoroughly and rigorously benchmark physical attacks with fairness under controlled physical dynamics and cross-domain transformation. This resolves the problem of capturing identical adversarial images that cannot be achieved in the real world. Our benchmark includes 20 physical attack methods, 48 object detectors, comprehensive physical dynamics, and evaluation metrics. We also provide end-to-end pipelines for dataset generation, detection, evaluation, and further analysis. In addition, we perform 8064 groups of evaluation based on our benchmark, which includes both overall evaluation and further detailed ablation studies for controlled physical dynamics. Through these experiments, we provide in-depth analyses of physical attack performance and physical adversarial robustness, draw valuable observations, and discuss potential directions for future research. Codebase: https://github.com/JiaweiLian/Benchmarking_Physical_Attack
URLs: https://github.com/JiaweiLian/Benchmarking_Physical_Attack
Authors: Shuo Wang, Yongcai Wang, Zhimin Xu, Yongyu Guo, Wanting Li, Zhe Huang, Xuewei Bai, Deying Li
Abstract: For interacting with mobile objects in unfamiliar environments, simultaneously locating, mapping, and tracking the 3D poses of multiple objects are crucially required. This paper proposes a Tracklet Graph and Query Graph-based framework, i.e., GSLAMOT, to address this challenge. GSLAMOT utilizes camera and LiDAR multimodal information as inputs and divides the representation of the dynamic scene into a semantic map for representing the static environment, a trajectory of the ego-agent, and an online maintained Tracklet Graph (TG) for tracking and predicting the 3D poses of the detected mobile objects. A Query Graph (QG) is constructed in each frame by object detection to query and update TG. For accurate object association, a Multi-criteria Star Graph Association (MSGA) method is proposed to find matched objects between the detections in QG and the predicted tracklets in TG. Then, an Object-centric Graph Optimization (OGO) method is proposed to simultaneously optimize the TG, the semantic map, and the agent trajectory. It triangulates the detected objects into the map to enrich the map's semantic information. We address the efficiency issues to handle the three tightly coupled tasks in parallel. Experiments are conducted on KITTI, Waymo, and an emulated Traffic Congestion dataset that highlights challenging scenarios. Experiments show that GSLAMOT enables accurate crowded object tracking while conducting SLAM accurately in challenging scenarios, demonstrating more excellent performances than the state-of-the-art methods. The code and dataset are at https://gslamot.github.io.
Authors: Xueying Gu, Qiong Wu, Pingyi Fan, Qiang Fan, Nan Cheng, Wen Chen, Khaled B. Letaief
Abstract: In the Internet of Vehicles (IoV), Federated Learning (FL) provides a privacy-preserving solution by aggregating local models without sharing data. Traditional supervised learning requires image data with labels, but data labeling involves significant manual effort. Federated Self-Supervised Learning (FSSL) utilizes Self-Supervised Learning (SSL) for local training in FL, eliminating the need for labels while protecting privacy. Compared to other SSL methods, Momentum Contrast (MoCo) reduces the demand for computing resources and storage space by creating a dictionary. However, using MoCo in FSSL requires uploading the local dictionary from vehicles to Base Station (BS), which poses a risk of privacy leakage. Simplified Contrast (SimCo) addresses the privacy leakage issue in MoCo-based FSSL by using dual temperature instead of a dictionary to control sample distribution. Additionally, considering the negative impact of motion blur on model aggregation, and based on SimCo, we propose a motion blur-resistant FSSL method, referred to as BFSSL. Furthermore, we address energy consumption and delay in the BFSSL process by proposing a Deep Reinforcement Learning (DRL)-based resource allocation scheme, called DRL-BFSSL. In this scheme, BS allocates the Central Processing Unit (CPU) frequency and transmission power of vehicles to minimize energy consumption and latency, while aggregating received models based on the motion blur level. Simulation results validate the effectiveness of our proposed aggregation and resource allocation methods.
Authors: Junlin Chen, Chengcheng Xu, Yangfan Xu, Jian Yang, Jun Li, Zhiping Shi
Abstract: In recent years, video action recognition, as a fundamental task in the field of video understanding, has been deeply explored by numerous researchers.Most traditional video action recognition methods typically involve converting videos into three-dimensional data that encapsulates both spatial and temporal information, subsequently leveraging prevalent image understanding models to model and analyze these data. However,these methods have significant drawbacks. Firstly, when delving into video action recognition tasks, image understanding models often need to be adapted accordingly in terms of model architecture and preprocessing for these spatiotemporal tasks; Secondly, dealing with high-dimensional data often poses greater challenges and incurs higher time costs compared to its lower-dimensional counterparts.To bridge the gap between image-understanding and video-understanding tasks while simplifying the complexity of video comprehension, we introduce a novel video representation architecture, Flatten, which serves as a plug-and-play module that can be seamlessly integrated into any image-understanding network for efficient and effective 3D temporal data modeling.Specifically, by applying specific flattening operations (e.g., row-major transform), 3D spatiotemporal data is transformed into 2D spatial information, and then ordinary image understanding models are used to capture temporal dynamic and spatial semantic information, which in turn accomplishes effective and efficient video action recognition. Extensive experiments on commonly used datasets (Kinetics-400, Something-Something v2, and HMDB-51) and three classical image classification models (Uniformer, SwinV2, and ResNet), have demonstrated that embedding Flatten provides a significant performance improvements over original model.
Authors: Zhaoli Deng, Kaibin Zhou, Fanyi Wang, Zhenpeng Mi
Abstract: With the wide application of diffusion model, the high cost of inference resources has became an important bottleneck for its universal application. Controllable generation, such as ControlNet, is one of the key research directions of diffusion model, and the research related to inference acceleration and model compression is more important. In order to solve this problem, this paper proposes a modal reparameterization method, RepControlNet, to realize the controllable generation of diffusion models without increasing computation. In the training process, RepControlNet uses the adapter to modulate the modal information into the feature space, copy the CNN and MLP learnable layers of the original diffusion model as the modal network, and initialize these weights based on the original weights and coefficients. The training process only optimizes the parameters of the modal network. In the inference process, the weights of the neutralization original diffusion model in the modal network are reparameterized, which can be compared with or even surpass the methods such as ControlNet, which use additional parameters and computational quantities, without increasing the number of parameters. We have carried out a large number of experiments on both SD1.5 and SDXL, and the experimental results show the effectiveness and efficiency of the proposed RepControlNet.
Authors: Xin Lin, Yuyan Zhou, Jingtong Yue, Chao Ren, Kelvin C. K. Chan, Lu Qi, Ming-Hsuan Yang
Abstract: Unsupervised restoration approaches based on generative adversarial networks (GANs) offer a promising solution without requiring paired datasets. Yet, these GAN-based approaches struggle to surpass the performance of conventional unsupervised GAN-based frameworks without significantly modifying model structures or increasing the computational complexity. To address these issues, we propose a self-collaboration (SC) strategy for existing restoration models. This strategy utilizes information from the previous stage as feedback to guide subsequent stages, achieving significant performance improvement without increasing the framework's inference complexity. The SC strategy comprises a prompt learning (PL) module and a restorer ($Res$). It iteratively replaces the previous less powerful fixed restorer $\overline{Res}$ in the PL module with a more powerful $Res$. The enhanced PL module generates better pseudo-degraded/clean image pairs, leading to a more powerful $Res$ for the next iteration. Our SC can significantly improve the $Res$'s performance by over 1.5 dB without adding extra parameters or computational complexity during inference. Meanwhile, existing self-ensemble (SE) and our SC strategies enhance the performance of pre-trained restorers from different perspectives. As SE increases computational complexity during inference, we propose a re-boosting module to the SC (Reb-SC) to improve the SC strategy further by incorporating SE into SC without increasing inference time. This approach further enhances the restorer's performance by approximately 0.3 dB. Extensive experimental results on restoration tasks demonstrate that the proposed model performs favorably against existing state-of-the-art unsupervised restoration methods. Source code and trained models are publicly available at: \url{https://github.com/linxin0/RSCP2GAN}.
Authors: Zhaoli Deng, Wen Liu, Fanyi Wang, Junkang Zhang, Fan Chen, Wendong Zhang, Zhenpeng Mi
Abstract: Portrait Fidelity Generation is a prominent research area in generative models, with a primary focus on enhancing both controllability and fidelity. Current methods face challenges in generating high-fidelity portrait results when faces occupy a small portion of the image with a low resolution, especially in multi-person group photo settings. To tackle these issues, we propose a systematic solution called MagicID, based on a self-constructed million-level multi-modal dataset named IDZoom. MagicID consists of Multi-Mode Fusion training strategy (MMF) and DDIM Inversion based ID Restoration inference framework (DIIR). During training, MMF iteratively uses the skeleton and landmark modalities from IDZoom as conditional guidance. By introducing the Clone Face Tuning in training stage and Mask Guided Multi-ID Cross Attention (MGMICA) in inference stage, explicit constraints on face positional features are achieved for multi-ID group photo generation. The DIIR aims to address the issue of artifacts. The DDIM Inversion is used in conjunction with face landmarks, global and local face features to achieve face restoration while keeping the background unchanged. Additionally, DIIR is plug-and-play and can be applied to any diffusion-based portrait generation method. To validate the effectiveness of MagicID, we conducted extensive comparative and ablation experiments. The experimental results demonstrate that MagicID has significant advantages in both subjective and objective metrics, and achieves controllable generation in multi-person scenarios.
Authors: Zinan Xiong, Shuijiao Chen, Yizhe Zhang, Yu Cao, Benyuan Liu, Xiaowei Liu
Abstract: Atrophic gastritis is a significant risk factor for developing gastric cancer. The incorporation of machine learning algorithms can efficiently elevate the possibility of accurately detecting atrophic gastritis. Nevertheless, when the trained model is applied in real-life circumstances, its output is often not consistently reliable. In this paper, we propose Adaptify, an adaptation scheme in which the model assimilates knowledge from its own classification decisions. Our proposed approach includes keeping the primary model constant, while simultaneously running and updating the auxiliary model. By integrating the knowledge gleaned by the auxiliary model into the primary model and merging their outputs, we have observed a notable improvement in output stability and consistency compared to relying solely on either the main model or the auxiliary model.
Authors: Daniel Kathein, Byron Hernandez, Henry Medeiros
Abstract: Multi-camera Association (MCA) is the task of identifying objects and individuals across camera views and is an active research topic, given its numerous applications across robotics, surveillance, and agriculture. We investigate a novel multi-camera multi-target association algorithm based on dense pixel correspondence estimation with a Transformer-based architecture and underlying detection-based masking. After the algorithm generates a set of corresponding keypoints and their respective confidence levels between every pair of detections in the camera views are computed, an affinity matrix is determined containing the probabilities of matches between each pair. Finally, the Hungarian algorithm is applied to generate an optimal assignment matrix with all the predicted associations between the camera views. Our method is evaluated on the WILDTRACK Seven-Camera HD Dataset, a high-resolution dataset containing footage of walking pedestrians as well as precise annotations and camera calibrations. Our results conclude that the algorithm performs exceptionally well associating pedestrians on camera pairs that are positioned close to each other and observe the scene from similar perspectives. On camera pairs with orientations that are drastically different in distance or angle, there is still significant room for improvement.
Authors: Chien-Yao Wang, Hong-Yuan Mark Liao
Abstract: This is a comprehensive review of the YOLO series of systems. Different from previous literature surveys, this review article re-examines the characteristics of the YOLO series from the latest technical point of view. At the same time, we also analyzed how the YOLO series continued to influence and promote real-time computer vision-related research and led to the subsequent development of computer vision and language models.We take a closer look at how the methods proposed by the YOLO series in the past ten years have affected the development of subsequent technologies and show the applications of YOLO in various fields. We hope this article can play a good guiding role in subsequent real-time computer vision development.
Authors: Hao Ai, Lin Wang
Abstract: 360 cameras capture the entire surrounding environment with a large FoV, exhibiting comprehensive visual information to directly infer the 3D structures, e.g., depth and surface normal, and semantic information simultaneously. Existing works predominantly specialize in a single task, leaving multi-task learning of 3D geometry and semantics largely unexplored. Achieving such an objective is, however, challenging due to: 1) inherent spherical distortion of planar equirectangular projection (ERP) and insufficient global perception induced by 360 image's ultra-wide FoV; 2) non-trivial progress in effectively merging geometry and semantics among different tasks to achieve mutual benefits. In this paper, we propose a novel end-to-end multi-task learning framework, named Elite360M, capable of inferring 3D structures via depth and surface normal estimation, and semantics via semantic segmentation simultaneously. Our key idea is to build a representation with strong global perception and less distortion while exploring the inter- and cross-task relationships between geometry and semantics. We incorporate the distortion-free and spatially continuous icosahedron projection (ICOSAP) points and combine them with ERP to enhance global perception. With a negligible cost, a Bi-projection Bi-attention Fusion module is thus designed to capture the semantic- and distance-aware dependencies between each pixel of the region-aware ERP feature and the ICOSAP point feature set. Moreover, we propose a novel Cross-task Collaboration module to explicitly extract task-specific geometric and semantic information from the learned representation to achieve preliminary predictions. It then integrates the spatial contextual information among tasks to realize cross-task fusion. Extensive experiments demonstrate the effectiveness and efficacy of Elite360M.
Authors: Dongze Li, Kang Zhao, Wei Wang, Yifeng Ma, Bo Peng, Yingya Zhang, Jing Dong
Abstract: Talking head synthesis is a practical technique with wide applications. Current Neural Radiance Field (NeRF) based approaches have shown their superiority on driving one-shot talking heads with videos or signals regressed from audio. However, most of them failed to take the audio as driven information directly, unable to enjoy the flexibility and availability of speech. Since mapping audio signals to face deformation is non-trivial, we design a Single-Shot Speech-Driven Neural Radiance Field (S^3D-NeRF) method in this paper to tackle the following three difficulties: learning a representative appearance feature for each identity, modeling motion of different face regions with audio, and keeping the temporal consistency of the lip area. To this end, we introduce a Hierarchical Facial Appearance Encoder to learn multi-scale representations for catching the appearance of different speakers, and elaborate a Cross-modal Facial Deformation Field to perform speech animation according to the relationship between the audio signal and different face regions. Moreover, to enhance the temporal consistency of the important lip area, we introduce a lip-sync discriminator to penalize the out-of-sync audio-visual sequences. Extensive experiments have shown that our S^3D-NeRF surpasses previous arts on both video fidelity and audio-lip synchronization.
Authors: Haoyun Qin, Jian Lin, Hanyuan Liu, Xueting Liu, Chengze Li
Abstract: Assistive drawing aims to facilitate the creative process by providing intelligent guidance to artists. Existing solutions often fail to effectively model intricate stroke details or adequately address the temporal aspects of drawing. We introduce hyperstroke, a novel stroke representation designed to capture precise fine stroke details, including RGB appearance and alpha-channel opacity. Using a Vector Quantization approach, hyperstroke learns compact tokenized representations of strokes from real-life drawing videos of artistic drawing. With hyperstroke, we propose to model assistive drawing via a transformer-based architecture, to enable intuitive and user-friendly drawing applications, which are experimented in our exploratory evaluation.
Authors: Jihwan Kim, Jaehyun Choi, Yerim Jeon, Jae-Pil Heo
Abstract: Temporal action detection (TAD) is challenging, yet fundamental for real-world video applications. Large temporal scale variation of actions is one of the most primary difficulties in TAD. Naturally, multi-scale features have potential in localizing actions of diverse lengths as widely used in object detection. Nevertheless, unlike objects in images, actions have more ambiguity in their boundaries. That is, small neighboring objects are not considered as a large one while short adjoining actions can be misunderstood as a long one. In the coarse-to-fine feature pyramid via pooling, these vague action boundaries can fade out, which we call 'vanishing boundary problem'. To this end, we propose Boundary-Recovering Network (BRN) to address the vanishing boundary problem. BRN constructs scale-time features by introducing a new axis called scale dimension by interpolating multi-scale features to the same temporal length. On top of scale-time features, scale-time blocks learn to exchange features across scale levels, which can effectively settle down the issue. Our extensive experiments demonstrate that our model outperforms the state-of-the-art on the two challenging benchmarks, ActivityNet-v1.3 and THUMOS14, with remarkably reduced degree of the vanishing boundary problem.
Authors: Shanaka Ramesh Gunasekara, Wanqing Li, Jack Yang, Philip Ogunbona
Abstract: In skeleton-based human action recognition, temporal pooling is a critical step for capturing spatiotemporal relationship of joint dynamics. Conventional pooling methods overlook the preservation of motion information and treat each frame equally. However, in an action sequence, only a few segments of frames carry discriminative information related to the action. This paper presents a novel Joint Motion Adaptive Temporal Pooling (JMAP) method for improving skeleton-based action recognition. Two variants of JMAP, frame-wise pooling and joint-wise pooling, are introduced. The efficacy of JMAP has been validated through experiments on the popular NTU RGB+D 120 and PKU-MMD datasets.
Authors: Anusree P. S., Bikram Keshari Parida, Seong Yong Moon, Wonsang You
Abstract: Cone Beam Computed Tomography (CBCT) and Panoramic X-rays are the most commonly used imaging modalities in dental health care. CBCT can produce three-dimensional views of a patient's head, providing clinicians with better diagnostic capability, whereas Panoramic X-ray can capture the entire maxillofacial region in a single image. If the CBCT is already available, it can be beneficial to synthesize a Panoramic X-ray, thereby avoiding an immediate additional scan and extra radiation exposure. Existing methods focus on delineating an approximate dental arch and creating orthogonal projections along this arch. However, no golden standard is available for such dental arch extractions, and this choice can affect the quality of synthesized X-rays. To avoid such issues, we propose a novel method for synthesizing Panoramic X-rays from diverse head CBCTs, employing a simulated projection geometry and dynamic rotation centers. Our method effectively synthesized panoramic views from CBCT, even for patients with missing or nonexistent teeth and in the presence of severe metal implants. Our results demonstrate that this method can generate high-quality panoramic images irrespective of the CBCT scanner geometry.
Authors: Zhaoxuan Zhu, Chulong Chen, Bo Yang
Abstract: Automotive Multiple-Input Multiple-Output (MIMO) radars have gained significant traction in Advanced Driver Assistance Systems (ADAS) and Autonomous Vehicles (AV) due to their cost-effectiveness, resilience to challenging operating conditions, and extended detection range. To fully leverage the advantages of MIMO radars, it is crucial to develop an Angle of Arrival (AOA) algorithm that delivers high performance with reasonable computational workload. This work introduces AAETR (Angle of Arrival Estimation with TRansformer) for high performance gridless AOA estimation. Comprehensive evaluations across various signal-to-noise ratios (SNRs) and multi-target scenarios demonstrate AAETR's superior performance compared to super resolution AOA algorithms such as Iterative Adaptive Approach (IAA). The proposed architecture features efficient, scalable, sparse and gridless angle-finding capability, overcoming the issues of high computational cost and straddling loss in SNR associated with grid-based IAA. AAETR requires fewer tunable hyper-parameters and is end-to-end trainable in a deep learning radar perception pipeline. When trained on large-scale simulated datasets then evaluated on real dataset, AAETR exhibits remarkable zero-shot sim-to-real transferability and emergent sidelobe suppression capability. This highlights the effectiveness of the proposed approach and its potential as a drop-in module in practical systems.
Authors: Taharim Rahman Anon, Jakaria Islam Emon
Abstract: As artificial intelligence progresses, the task of distinguishing between real and AI-generated images is increasingly complicated by sophisticated generative models. This paper presents a novel detection framework adept at robustly identifying images produced by cutting-edge generative AI models, such as DALL-E 3, MidJourney, and Stable Diffusion 3. We introduce a comprehensive dataset, tailored to include images from these advanced generators, which serves as the foundation for extensive evaluation. we propose a classification system that integrates semantic image embeddings with a traditional Multilayer Perceptron (MLP). This baseline system is designed to effectively differentiate between real and AI-generated images under various challenging conditions. Enhancing this approach, we introduce a hybrid architecture that combines Kolmogorov-Arnold Networks (KAN) with the MLP. This hybrid model leverages the adaptive, high-resolution feature transformation capabilities of KAN, enabling our system to capture and analyze complex patterns in AI-generated images that are typically overlooked by conventional models. In out-of-distribution testing, our proposed model consistently outperformed the standard MLP across three out of distribution test datasets, demonstrating superior performance and robustness in classifying real images from AI-generated images with impressive F1 scores.
Authors: Ziyu Yao, Xuxin Cheng, Zhiqi Huang
Abstract: Talking head generation is a significant research topic that still faces numerous challenges. Previous works often adopt generative adversarial networks or regression models, which are plagued by generation quality and average facial shape problem. Although diffusion models show impressive generative ability, their exploration in talking head generation remains unsatisfactory. This is because they either solely use the diffusion model to obtain an intermediate representation and then employ another pre-trained renderer, or they overlook the feature decoupling of complex facial details, such as expressions, head poses and appearance textures. Therefore, we propose a Facial Decoupled Diffusion model for Talking head generation called FD2Talk, which fully leverages the advantages of diffusion models and decouples the complex facial details through multi-stages. Specifically, we separate facial details into motion and appearance. In the initial phase, we design the Diffusion Transformer to accurately predict motion coefficients from raw audio. These motions are highly decoupled from appearance, making them easier for the network to learn compared to high-dimensional RGB images. Subsequently, in the second phase, we encode the reference image to capture appearance textures. The predicted facial and head motions and encoded appearance then serve as the conditions for the Diffusion UNet, guiding the frame generation. Benefiting from decoupling facial details and fully leveraging diffusion models, extensive experiments substantiate that our approach excels in enhancing image quality and generating more accurate and diverse results compared to previous state-of-the-art methods.
Authors: Yang Li, Jianing Deng, Chong Zhong, Danjuan Yang, Meiyan Li, A. H. Welsh, Aiyi Liu, Xingtao Zhou, Catherine C. Liu, Bo Fu
Abstract: Myopia screening using cutting-edge ultra-widefield (UWF) fundus imaging and joint modeling of multiple discrete and continuous clinical scores presents a promising new paradigm for multi-task problems in Ophthalmology. The bi-channel framework that arises from the Ophthalmic phenomenon of ``interocular asymmetries'' of both eyes (OU) calls for new employment on the SOTA transformer-based models. However, the application of copula models for multiple mixed discrete-continuous labels on deep learning (DL) is challenging. Moreover, the application of advanced large transformer-based models to small medical datasets is challenging due to overfitting and computational resource constraints. To resolve these challenges, we propose OU-CoViT: a novel Copula-Enhanced Bi-Channel Multi-Task Vision Transformers with Dual Adaptation for OU-UWF images, which can i) incorporate conditional correlation information across multiple discrete and continuous labels within a deep learning framework (by deriving the closed form of a novel Copula Loss); ii) take OU inputs subject to both high correlation and interocular asymmetries using a bi-channel model with dual adaptation; and iii) enable the adaptation of large vision transformer (ViT) models to small medical datasets. Solid experiments demonstrate that OU-CoViT significantly improves prediction performance compared to single-channel baseline models with empirical loss. Furthermore, the novel architecture of OU-CoViT allows generalizability and extensions of our dual adaptation and Copula Loss to various ViT variants and large DL models on small medical datasets. Our approach opens up new possibilities for joint modeling of heterogeneous multi-channel input and mixed discrete-continuous clinical scores in medical practices and has the potential to advance AI-assisted clinical decision-making in various medical domains beyond Ophthalmology.
Authors: Chao Xu, Mingze Sun, Zhi-Qi Cheng, Fei Wang, Yang Liu, Baigui Sun, Ruqi Huang, Alexander Hauptmann
Abstract: In this paper, we propose a novel framework, Combo, for harmonious co-speech holistic 3D human motion generation and efficient customizable adaption. In particular, we identify that one fundamental challenge as the multiple-input-multiple-output (MIMO) nature of the generative model of interest. More concretely, on the input end, the model typically consumes both speech signals and character guidance (e.g., identity and emotion), which not only poses challenge on learning capacity but also hinders further adaptation to varying guidance; on the output end, holistic human motions mainly consist of facial expressions and body movements, which are inherently correlated but non-trivial to coordinate in current data-driven generation process. In response to the above challenge, we propose tailored designs to both ends. For the former, we propose to pre-train on data regarding a fixed identity with neutral emotion, and defer the incorporation of customizable conditions (identity and emotion) to fine-tuning stage, which is boosted by our novel X-Adapter for parameter-efficient fine-tuning. For the latter, we propose a simple yet effective transformer design, DU-Trans, which first divides into two branches to learn individual features of face expression and body movements, and then unites those to learn a joint bi-directional distribution and directly predicts combined coefficients. Evaluated on BEAT2 and SHOW datasets, Combo is highly effective in generating high-quality motions but also efficient in transferring identity and emotion. Project website: \href{https://xc-csc101.github.io/combo/}{Combo}.
Authors: Xinjie Jiang, Chenxi Zheng, Xuemiao Xu, Bangzhen Liu, Weiying Zheng, Huaidong Zhang, Shengfeng He
Abstract: Video Visual Relation Detection (VidVRD) focuses on understanding how entities interact over time and space in videos, a key step for gaining deeper insights into video scenes beyond basic visual tasks. Traditional methods for VidVRD, challenged by its complexity, typically split the task into two parts: one for identifying what relation categories are present and another for determining their temporal boundaries. This split overlooks the inherent connection between these elements. Addressing the need to recognize entity pairs' spatiotemporal interactions across a range of durations, we propose VrdONE, a streamlined yet efficacious one-stage model. VrdONE combines the features of subjects and objects, turning predicate detection into 1D instance segmentation on their combined representations. This setup allows for both relation category identification and binary mask generation in one go, eliminating the need for extra steps like proposal generation or post-processing. VrdONE facilitates the interaction of features across various frames, adeptly capturing both short-lived and enduring relations. Additionally, we introduce the Subject-Object Synergy (SOS) module, enhancing how subjects and objects perceive each other before combining. VrdONE achieves state-of-the-art performances on the VidOR benchmark and ImageNet-VidVRD, showcasing its superior capability in discerning relations across different temporal scales. The code is available at \textcolor[RGB]{228,58,136}{\href{https://github.com/lucaspk512/vrdone}{https://github.com/lucaspk512/vrdone}}.
URLs: https://github.com/lucaspk512/vrdone, https://github.com/lucaspk512/vrdone
Authors: Chen Long-fei, Subramanian Ramamoorthy, Robert B Fisher
Abstract: Vision-based motion estimation methods show promise in accurately and unobtrusively estimating human body motion for healthcare purposes. However, these methods are not specifically designed for healthcare purposes and face challenges in real-world applications. Human pose estimation methods often lack the accuracy needed for detecting fine-grained, subtle body movements, while optical flow-based methods struggle with poor lighting conditions and unseen real-world data. These issues result in human body motion estimation errors, particularly during critical medical situations where the body is motionless, such as during unconsciousness. To address these challenges and improve the accuracy of human body motion estimation for healthcare purposes, we propose the OPPH operator designed to enhance current vision-based motion estimation methods. This operator, which considers human body movement and noise properties, functions as a multi-stage filter. Results tested on two real-world and one synthetic human motion dataset demonstrate that the operator effectively removes real-world noise, significantly enhances the detection of motionless states, maintains the accuracy of estimating active body movements, and maintains long-term body movement trends. This method could be beneficial for analyzing both critical medical events and chronic medical conditions.
Authors: Litingyu Wang (University of Electronic Science and Technology of China, Chengdu, China), Yijie Qu (University of Electronic Science and Technology of China, Chengdu, China), Xiangde Luo (University of Electronic Science and Technology of China, Chengdu, China, ShangAI Laboratory, Shanghai, China), Wenjun Liao (University of Electronic Science and Technology of China, Chengdu, China, Department of Radiation Oncology, Sichuan Cancer Hospital & Institute, Sichuan Cancer Center, Chengdu, China), Shichuan Zhang (University of Electronic Science and Technology of China, Chengdu, China, Department of Radiation Oncology, Sichuan Cancer Hospital & Institute, Sichuan Cancer Center, Chengdu, China), Guotai Wang (University of Electronic Science and Technology of China, Chengdu, China, ShangAI Laboratory, Shanghai, China)
Abstract: Assessing the presence of potentially malignant lymph nodes aids in estimating cancer progression, and identifying surrounding benign lymph nodes can assist in determining potential metastatic pathways for cancer. For quantitative analysis, automatic segmentation of lymph nodes is crucial. However, due to the labor-intensive and time-consuming manual annotation process required for a large number of lymph nodes, it is more practical to annotate only a subset of the lymph node instances to reduce annotation costs. In this study, we propose a pre-trained Dual-Branch network with Dynamically Mixed Pseudo label (DBDMP) to learn from partial instance annotations for lymph nodes segmentation. To obtain reliable pseudo labels for lymph nodes that are not annotated, we employ a dual-decoder network to generate different outputs that are then dynamically mixed. We integrate the original weak partial annotations with the mixed pseudo labels to supervise the network. To further leverage the extensive amount of unannotated voxels, we apply a self-supervised pre-training strategy to enhance the model's feature extraction capability. Experiments on the mediastinal Lymph Node Quantification (LNQ) dataset demonstrate that our method, compared to directly learning from partial instance annotations, significantly improves the Dice Similarity Coefficient (DSC) from 11.04% to 54.10% and reduces the Average Symmetric Surface Distance (ASSD) from 20.83 $mm$ to 8.72 $mm$. The code is available at https://github.com/WltyBY/LNQ2023_training_code.git
Authors: Muhammad Rameez Ur Rahman, Jhony H. Giraldo, Indro Spinelli, St\'ephane Lathuili\`ere, Fabio Galasso
Abstract: Event cameras, known for low-latency operation and superior performance in challenging lighting conditions, are suitable for sensitive computer vision tasks such as semantic segmentation in autonomous driving. However, challenges arise due to limited event-based data and the absence of large-scale segmentation benchmarks. Current works are confined to closed-set semantic segmentation, limiting their adaptability to other applications. In this paper, we introduce OVOSE, the first Open-Vocabulary Semantic Segmentation algorithm for Event cameras. OVOSE leverages synthetic event data and knowledge distillation from a pre-trained image-based foundation model to an event-based counterpart, effectively preserving spatial context and transferring open-vocabulary semantic segmentation capabilities. We evaluate the performance of OVOSE on two driving semantic segmentation datasets DDD17, and DSEC-Semantic, comparing it with existing conventional image open-vocabulary models adapted for event-based data. Similarly, we compare OVOSE with state-of-the-art methods designed for closed-set settings in unsupervised domain adaptation for event-based semantic segmentation. OVOSE demonstrates superior performance, showcasing its potential for real-world applications. The code is available at https://github.com/ram95d/OVOSE.
Authors: Mahrukh Siddiqui, Shahzaib Iqbal, Bandar AlShammari, Bandar Alhaqbani, Tariq M. Khan, Imran Razzak
Abstract: Compared to contact fingerprint images, contactless fingerprint images exhibit four distinct characteristics: (1) they contain less noise; (2) they have fewer discontinuities in ridge patterns; (3) the ridge-valley pattern is less distinct; and (4) they pose an interoperability problem, as they lack the elastic deformation caused by pressing the finger against the capture device. These properties present significant challenges for the enhancement of contactless fingerprint images. In this study, we propose a novel contactless fingerprint identification solution that enhances the accuracy of minutiae detection through improved frequency estimation and a new region-quality-based minutia extraction algorithm. In addition, we introduce an efficient and highly accurate minutiae-based encoding and matching algorithm. We validate the effectiveness of our approach through extensive experimental testing. Our method achieves a minimum Equal Error Rate (EER) of 2.84\% on the PolyU contactless fingerprint dataset, demonstrating its superior performance compared to existing state-of-the-art techniques. The proposed fingerprint identification method exhibits notable precision and resilience, proving to be an effective and feasible solution for contactless fingerprint-based identification systems.
Authors: Kaiwen Wang, Yinzhe Shen, Martin Lauer
Abstract: Object detectors encounter challenges in handling domain shifts. Cutting-edge domain adaptive object detection methods use the teacher-student framework and domain adversarial learning to generate domain-invariant pseudo-labels for self-training. However, the pseudo-labels generated by the teacher model tend to be biased towards the majority class and often mistakenly include overconfident false positives and underconfident false negatives. We reveal that pseudo-labels vulnerable to adversarial attacks are more likely to be low-quality. To address this, we propose a simple yet effective framework named Adversarial Attacked Teacher (AAT) to improve the quality of pseudo-labels. Specifically, we apply adversarial attacks to the teacher model, prompting it to generate adversarial pseudo-labels to correct bias, suppress overconfidence, and encourage underconfident proposals. An adaptive pseudo-label regularization is introduced to emphasize the influence of pseudo-labels with high certainty and reduce the negative impacts of uncertain predictions. Moreover, robust minority objects verified by pseudo-label regularization are oversampled to minimize dataset imbalance without introducing false positives. Extensive experiments conducted on various datasets demonstrate that AAT achieves superior performance, reaching 52.6 mAP on Clipart1k, surpassing the previous state-of-the-art by 6.7%.
Authors: Kaicheng Yang, Tiancheng Gu, Xiang An, Haiqiang Jiang, Xiangzi Dai, Ziyong Feng, Weidong Cai, Jiankang Deng
Abstract: Contrastive Language-Image Pre-training (CLIP) has achieved excellent performance over a wide range of tasks. However, the effectiveness of CLIP heavily relies on a substantial corpus of pre-training data, resulting in notable consumption of computational resources. Although knowledge distillation has been widely applied in single modality models, how to efficiently expand knowledge distillation to vision-language foundation models with extensive data remains relatively unexplored. In this paper, we introduce CLIP-CID, a novel distillation mechanism that effectively transfers knowledge from a large vision-language foundation model to a smaller model. We initially propose a simple but efficient image semantic balance method to reduce transfer learning bias and improve distillation efficiency. This method filters out 43.7% of image-text pairs from the LAION400M while maintaining superior performance. After that, we leverage cluster-instance discrimination to facilitate knowledge transfer from the teacher model to the student model, thereby empowering the student model to acquire a holistic semantic comprehension of the pre-training data. Experimental results demonstrate that CLIP-CID achieves state-of-the-art performance on various downstream tasks including linear probe and zero-shot classification.
Authors: Xin Liu, Weijia Zhang, Min-Ling Zhang
Abstract: Although attention-based multi-instance learning algorithms have achieved impressive performances on slide-level whole slide image (WSI) classification tasks, they are prone to mistakenly focus on irrelevant patterns such as staining conditions and tissue morphology, leading to incorrect patch-level predictions and unreliable interpretability. Moreover, these attention-based MIL algorithms tend to focus on salient instances and struggle to recognize hard-to-classify instances. In this paper, we first demonstrate that attention-based WSI classification methods do not adhere to the standard MIL assumptions. From the standard MIL assumptions, we propose a surprisingly simple yet effective instance-based MIL method for WSI classification (FocusMIL) based on max-pooling and forward amortized variational inference. We argue that synergizing the standard MIL assumption with variational inference encourages the model to focus on tumour morphology instead of spurious correlations. Our experimental evaluations show that FocusMIL significantly outperforms the baselines in patch-level classification tasks on the Camelyon16 and TCGA-NSCLC benchmarks. Visualization results show that our method also achieves better classification boundaries for identifying hard instances and mitigates the effect of spurious correlations between bags and labels.
Authors: Victoria Clerico (George Mason Unviersity), Shay Snyder (George Mason Unviersity), Arya Lohia (George Mason Unviersity), Md Abdullah-Al Kaiser (University of Southern, California), Gregory Schwartz (Northwestern University), Akhilesh Jaiswal (University of Southern, California), Maryam Parsa (George Mason Unviersity)
Abstract: Dynamic Vision Sensors (DVS) have emerged as a revolutionary technology with a high temporal resolution that far surpasses RGB cameras. DVS technology draws biological inspiration from photoreceptors and the initial retinal synapse. Our research showcases the potential of additional retinal functionalities to extract visual features. We provide a domain-agnostic and efficient algorithm for ego-motion compensation based on Object Motion Sensitivity (OMS), one of the multiple robust features computed within the mammalian retina. We develop a framework based on experimental neuroscience that translates OMS' biological circuitry to a low-overhead algorithm. OMS processes DVS data from dynamic scenes to perform pixel-wise object motion segmentation. Using a real and a synthetic dataset, we highlight OMS' ability to differentiate object motion from ego-motion, bypassing the need for deep networks. This paper introduces a bio-inspired computer vision method that dramatically reduces the number of parameters by a factor of 1000 compared to prior works. Our work paves the way for robust, high-speed, and low-bandwidth decision-making for in-sensor computations.
Authors: Haoxin Yang, Xuemiao Xu, Cheng Xu, Huaidong Zhang, Jing Qin, Yi Wang, Pheng-Ann Heng, Shengfeng He
Abstract: Reversible face anonymization, unlike traditional face pixelization, seeks to replace sensitive identity information in facial images with synthesized alternatives, preserving privacy without sacrificing image clarity. Traditional methods, such as encoder-decoder networks, often result in significant loss of facial details due to their limited learning capacity. Additionally, relying on latent manipulation in pre-trained GANs can lead to changes in ID-irrelevant attributes, adversely affecting data utility due to GAN inversion inaccuracies. This paper introduces G\textsuperscript{2}Face, which leverages both generative and geometric priors to enhance identity manipulation, achieving high-quality reversible face anonymization without compromising data utility. We utilize a 3D face model to extract geometric information from the input face, integrating it with a pre-trained GAN-based decoder. This synergy of generative and geometric priors allows the decoder to produce realistic anonymized faces with consistent geometry. Moreover, multi-scale facial features are extracted from the original face and combined with the decoder using our novel identity-aware feature fusion blocks (IFF). This integration enables precise blending of the generated facial patterns with the original ID-irrelevant features, resulting in accurate identity manipulation. Extensive experiments demonstrate that our method outperforms existing state-of-the-art techniques in face anonymization and recovery, while preserving high data utility. Code is available at https://github.com/Harxis/G2Face.
Authors: Weijia Li, Jinhua Yu, Dairong Chen, Yi Lin, Runming Dong, Xiang Zhang, Conghui He, Haohuan Fu
Abstract: In this work, we propose a geometry-aware semi-supervised method for fine-grained building function recognition. This method leverages the geometric relationships between multi-source data to improve the accuracy of pseudo labels in semi-supervised learning, extending the task's scope and making it applicable to cross-categorization systems of building function recognition. Firstly, we design an online semi-supervised pre-training stage, which facilitates the precise acquisition of building facade location information in street-view images. In the second stage, we propose a geometry-aware coarse annotation generation module. This module effectively combines GIS data and street-view data based on the geometric relationships, improving the accuracy of pseudo annotations. In the third stage, we combine the newly generated coarse annotations with the existing labeled dataset to achieve fine-grained functional recognition of buildings across multiple cities at a large scale. Extensive experiments demonstrate that our proposed framework exhibits superior performance in fine-grained functional recognition of buildings. Within the same categorization system, it achieves improvements of 7.6% and 4.8% compared to fully-supervised methods and state-of-the-art semi-supervised methods, respectively. Additionally, our method also performs well in cross-city tasks, i.e., extending the model trained on OmniCity (New York) to new areas (i.e., Los Angeles and Boston). This study provides a novel solution for the fine-grained function recognition of large-scale buildings across multiple cities, offering essential data for understanding urban infrastructure planning, human activity patterns, and the interactions between humans and buildings.
Authors: Mingxiao Zheng, Yanpeng Qu, Changjing Shang, Longzhi Yang, Qiang Shen
Abstract: Unsupervised person re-identification (Re-ID) aims to learn a feature network with cross-camera retrieval capability in unlabelled datasets. Although the pseudo-label based methods have achieved great progress in Re-ID, their performance in the complex scenario still needs to sharpen up. In order to reduce potential misguidance, including feature bias, noise pseudo-labels and invalid hard samples, accumulated during the learning process, in this pa per, a confidence-guided clustering and contrastive learning (3C) framework is proposed for unsupervised person Re-ID. This 3C framework presents three confidence degrees. i) In the clustering stage, the confidence of the discrepancy between samples and clusters is proposed to implement a harmonic discrepancy clustering algorithm (HDC). ii) In the forward-propagation training stage, the confidence of the camera diversity of a cluster is evaluated via a novel camera information entropy (CIE). Then, the clusters with high CIE values will play leading roles in training the model. iii) In the back-propagation training stage, the confidence of the hard sample in each cluster is designed and further used in a confidence integrated harmonic discrepancy (CHD), to select the informative sample for updating the memory in contrastive learning. Extensive experiments on three popular Re-ID benchmarks demonstrate the superiority of the proposed framework. Particularly, the 3C framework achieves state-of-the-art results: 86.7%/94.7%, 45.3%/73.1% and 47.1%/90.6% in terms of mAP/Rank-1 accuracy on Market-1501, the com plex datasets MSMT17 and VeRi-776, respectively. Code is available at https://github.com/stone5265/3C-reid.
Authors: Tianyi Liu, Zhaorui Tan, Muyin Chen, Xi Yang, Haochuan Jiang, Kaizhu Huang
Abstract: Brain tumor segmentation is often based on multiple magnetic resonance imaging (MRI). However, in clinical practice, certain modalities of MRI may be missing, which presents a more difficult scenario. To cope with this challenge, Knowledge Distillation, Domain Adaption, and Shared Latent Space have emerged as commonly promising strategies. However, recent efforts typically overlook the modality gaps and thus fail to learn important invariant feature representations across different modalities. Such drawback consequently leads to limited performance for missing modality models. To ameliorate these problems, pre-trained models are used in natural visual segmentation tasks to minimize the gaps. However, promising pre-trained models are often unavailable in medical image segmentation tasks. Along this line, in this paper, we propose a novel paradigm that aligns latent features of involved modalities to a well-defined distribution anchor as the substitution of the pre-trained model}. As a major contribution, we prove that our novel training paradigm ensures a tight evidence lower bound, thus theoretically certifying its effectiveness. Extensive experiments on different backbones validate that the proposed paradigm can enable invariant feature representations and produce models with narrowed modality gaps. Models with our alignment paradigm show their superior performance on both BraTS2018 and BraTS2020 datasets.
Authors: Jun Wang, Yu Mao, Nan Guan, Chun Jason Xue
Abstract: Whole slide images (WSIs) are gigapixel-scale digital images of H\&E-stained tissue samples widely used in pathology. The substantial size and complexity of WSIs pose unique analytical challenges. Multiple Instance Learning (MIL) has emerged as a powerful approach for addressing these challenges, particularly in cancer classification and detection. This survey provides a comprehensive overview of the challenges and methodologies associated with applying MIL to WSI analysis, including attention mechanisms, pseudo-labeling, transformers, pooling functions, and graph neural networks. Additionally, it explores the potential of MIL in discovering cancer cell morphology, constructing interpretable machine learning models, and quantifying cancer grading. By summarizing the current challenges, methodologies, and potential applications of MIL in WSI analysis, this survey aims to inform researchers about the state of the field and inspire future research directions.
Authors: Yiran Song, Qianyu Zhou, Lizhuang Ma
Abstract: Surface defect detection is significant in industrial production. However, detecting defects with varying textures and anomaly classes during the test time is challenging. This arises due to the differences in data distributions between source and target domains. Collecting and annotating new data from the target domain and retraining the model is time-consuming and costly. In this paper, we propose a novel test-time adaptation surface-defect detection approach that adapts pre-trained models to new domains and classes during inference. Our approach involves two core ideas. Firstly, we introduce a supervisor to filter samples and select only those with high confidence to update the model. This ensures that the model is not excessively biased by incorrect data. Secondly, we propose the augmented mean prediction to generate robust pseudo labels and a dynamically-balancing loss to facilitate the model in effectively integrating classification and segmentation results to improve surface-defect detection accuracy. Our approach is real-time and does not require additional offline retraining. Experiments demonstrate it outperforms state-of-the-art techniques.
Authors: Wancheng Feng, Wanquan Feng, Dawei Huang, Jiaming Pei, Guangliang Cheng, Lukun Wang
Abstract: Stylization for visual content aims to add specific style patterns at the pixel level while preserving the original structural features. Compared with using predefined styles, stylization guided by reference style images is more challenging, where the main difficulty is to effectively separate style from structural elements. In this paper, we propose StyleBrush, a method that accurately captures styles from a reference image and ``brushes'' the extracted style onto other input visual content. Specifically, our architecture consists of two branches: ReferenceNet, which extracts style from the reference image, and Structure Guider, which extracts structural features from the input image, thus enabling image-guided stylization. We utilize LLM and T2I models to create a dataset comprising 100K high-quality style images, encompassing a diverse range of styles and contents with high aesthetic score. To construct training pairs, we crop different regions of the same training image. Experiments show that our approach achieves state-of-the-art results through both qualitative and quantitative analyses. We will release our code and dataset upon acceptance of the paper.
Authors: Chaofan Tao, Gukyeong Kwon, Varad Gunjal, Hao Yang, Zhaowei Cai, Yonatan Dukler, Ashwin Swaminathan, R. Manmatha, Colin Jon Taylor, Stefano Soatto
Abstract: We study the capability of Video-Language (VidL) models in understanding compositions between objects, attributes, actions and their relations. Composition understanding becomes particularly challenging for video data since the compositional relations rapidly change over time in videos. We first build a benchmark named AARO to evaluate composition understanding related to actions on top of spatial concepts. The benchmark is constructed by generating negative texts with incorrect action descriptions for a given video and the model is expected to pair a positive text with its corresponding video. Furthermore, we propose a training method called NAVERO which utilizes video-text data augmented with negative texts to enhance composition understanding. We also develop a negative-augmented visual-language matching loss which is used explicitly to benefit from the generated negative text. We compare NAVERO with other state-of-the-art methods in terms of compositional understanding as well as video-text retrieval performance. NAVERO achieves significant improvement over other methods for both video-language and image-language composition understanding, while maintaining strong performance on traditional text-video retrieval tasks.
Authors: Ying Zhao
Abstract: Recent advances in anomaly generation approaches alleviate the effect of data insufficiency on task of anomaly localization. While effective, most of them learn multiple large generative models on different datasets and cumbersome anomaly prediction models for different classes. To address the limitations, we propose a novel scalable framework, named AnomalyFactory, that unifies unsupervised anomaly generation and localization with same network architecture. It starts with a BootGenerator that combines structure of a target edge map and appearance of a reference color image with the guidance of a learned heatmap. Then, it proceeds with a FlareGenerator that receives supervision signals from the BootGenerator and reforms the heatmap to indicate anomaly locations in the generated image. Finally, it easily transforms the same network architecture to a BlazeDetector that localizes anomaly pixels with the learned heatmap by converting the anomaly images generated by the FlareGenerator to normal images. By manipulating the target edge maps and combining them with various reference images, AnomalyFactory generates authentic and diversity samples cross domains. Comprehensive experiments carried on 5 datasets, including MVTecAD, VisA, MVTecLOCO, MADSim and RealIAD, demonstrate that our approach is superior to competitors in generation capability and scalability.
Authors: Juan E. Tapia, Maximilian Russo, Christoph Busch
Abstract: Morphing Attack Detection (MAD) is a relevant topic that aims to detect attempts by unauthorised individuals to access a "valid" identity. One of the main scenarios is printing morphed images and submitting the respective print in a passport application process. Today, small datasets are available to train the MAD algorithm because of privacy concerns and the limitations resulting from the effort associated with the printing and scanning of images at large numbers. In order to improve the detection capabilities and spot such morphing attacks, it will be necessary to have a larger and more realistic dataset representing the passport application scenario with the diversity of devices and the resulting printed scanned or compressed images. Creating training data representing the diversity of attacks is a very demanding task because the training material is developed manually. This paper proposes two different methods based on transfer-transfer for automatically creating digital print/scan face images and using such images in the training of a Morphing Attack Detection algorithm. Our proposed method can reach an Equal Error Rate (EER) of 3.84% and 1.92% on the FRGC/FERET database when including our synthetic and texture-transfer print/scan with 600 dpi to handcrafted images, respectively.
Authors: Ushnish Sarkar, Archisman Chakraborti, Tapas Samanta, Sarbajit Pal, Amitabha Das
Abstract: This study presents a novel approach for enhancing American Sign Language (ASL) recognition using Graph Convolutional Networks (GCNs) integrated with successive residual connections. The method leverages the MediaPipe framework to extract key landmarks from each hand gesture, which are then used to construct graph representations. A robust preprocessing pipeline, including translational and scale normalization techniques, ensures consistency across the dataset. The constructed graphs are fed into a GCN-based neural architecture with residual connections to improve network stability. The architecture achieves state-of-the-art results, demonstrating superior generalization capabilities with a validation accuracy of 99.14%.
Authors: Boyang Li, Xinyi Ying, Ruojing Li, Yongxian Liu, Yangsi Shi, Miao Li
Abstract: In this paper, we briefly summarize the first competition on resource-limited infrared small target detection (namely, LimitIRSTD). This competition has two tracks, including weakly-supervised infrared small target detection (Track 1) and lightweight infrared small target detection (Track 2). 46 and 60 teams successfully registered and took part in Tracks 1 and Track 2, respectively. The top-performing methods and their results in each track are described with details. This competition inspires the community to explore the tough problems in the application of infrared small target detection, and ultimately promote the deployment of this technology under limited resource.
Authors: Chuangchuang Tan, Renshuai Tao, Huan Liu, Guanghua Gu, Baoyuan Wu, Yao Zhao, Yunchao Wei
Abstract: This work focuses on AIGC detection to develop universal detectors capable of identifying various types of forgery images. Recent studies have found large pre-trained models, such as CLIP, are effective for generalizable deepfake detection along with linear classifiers. However, two critical issues remain unresolved: 1) understanding why CLIP features are effective on deepfake detection through a linear classifier; and 2) exploring the detection potential of CLIP. In this study, we delve into the underlying mechanisms of CLIP's detection capabilities by decoding its detection features into text and performing word frequency analysis. Our finding indicates that CLIP detects deepfakes by recognizing similar concepts (Fig. \ref{fig:fig1} a). Building on this insight, we introduce Category Common Prompt CLIP, called C2P-CLIP, which integrates the category common prompt into the text encoder to inject category-related concepts into the image encoder, thereby enhancing detection performance (Fig. \ref{fig:fig1} b). Our method achieves a 12.41\% improvement in detection accuracy compared to the original CLIP, without introducing additional parameters during testing. Comprehensive experiments conducted on two widely-used datasets, encompassing 20 generation models, validate the efficacy of the proposed method, demonstrating state-of-the-art performance. The code is available at \url{https://github.com/chuangchuangtan/C2P-CLIP-DeepfakeDetection}
URLs: https://github.com/chuangchuangtan/C2P-CLIP-DeepfakeDetection
Authors: Eashan Adhikarla, Kai Zhang, John Nicholson, Brian D. Davison
Abstract: Low-light image enhancement remains a challenging task in computer vision, with existing state-of-the-art models often limited by hardware constraints and computational inefficiencies, particularly in handling high-resolution images. Recent foundation models, such as transformers and diffusion models, despite their efficacy in various domains, are limited in use on edge devices due to their computational complexity and slow inference times. We introduce ExpoMamba, a novel architecture that integrates components of the frequency state space within a modified U-Net, offering a blend of efficiency and effectiveness. This model is specifically optimized to address mixed exposure challenges, a common issue in low-light image enhancement, while ensuring computational efficiency. Our experiments demonstrate that ExpoMamba enhances low-light images up to 2-3x faster than traditional models with an inference time of 36.6 ms and achieves a PSNR improvement of approximately 15-20% over competing models, making it highly suitable for real-time image processing applications.
Authors: Haoyu Zhao, Hao Wang, Chen Yang, Wei Shen
Abstract: Recent advancements in human avatar synthesis have utilized radiance fields to reconstruct photo-realistic animatable human avatars. However, both NeRFs-based and 3DGS-based methods struggle with maintaining 3D consistency and exhibit suboptimal detail reconstruction, especially with sparse inputs. To address this challenge, we propose CHASE, which introduces supervision from intrinsic 3D consistency across poses and 3D geometry contrastive learning, achieving performance comparable with sparse inputs to that with full inputs. Following previous work, we first integrate a skeleton-driven rigid deformation and a non-rigid cloth dynamics deformation to coordinate the movements of individual Gaussians during animation, reconstructing basic avatar with coarse 3D consistency. To improve 3D consistency under sparse inputs, we design Dynamic Avatar Adjustment(DAA) to adjust deformed Gaussians based on a selected similar pose/image from the dataset. Minimizing the difference between the image rendered by adjusted Gaussians and the image with the similar pose serves as an additional form of supervision for avatar. Furthermore, we propose a 3D geometry contrastive learning strategy to maintain the 3D global consistency of generated avatars. Though CHASE is designed for sparse inputs, it surprisingly outperforms current SOTA methods \textbf{in both full and sparse settings} on the ZJU-MoCap and H36M datasets, demonstrating that our CHASE successfully maintains avatar's 3D consistency, hence improving rendering quality.
Authors: Haoyu Zhao, Chen Yang, Hao Wang, Xingyue Zhao, Wei Shen
Abstract: Reconstructing photo-realistic animatable human avatars from monocular videos remains challenging in computer vision and graphics. Recently, methods using 3D Gaussians to represent the human body have emerged, offering faster optimization and real-time rendering. However, due to ignoring the crucial role of human body semantic information which represents the intrinsic structure and connections within the human body, they fail to achieve fine-detail reconstruction of dynamic human avatars. To address this issue, we propose SG-GS, which uses semantics-embedded 3D Gaussians, skeleton-driven rigid deformation, and non-rigid cloth dynamics deformation to create photo-realistic animatable human avatars from monocular videos. We then design a Semantic Human-Body Annotator (SHA) which utilizes SMPL's semantic prior for efficient body part semantic labeling. The generated labels are used to guide the optimization of Gaussian semantic attributes. To address the limited receptive field of point-level MLPs for local features, we also propose a 3D network that integrates geometric and semantic associations for human avatar deformation. We further implement three key strategies to enhance the semantic accuracy of 3D Gaussians and rendering quality: semantic projection with 2D regularization, semantic-guided density regularization and semantic-aware regularization with neighborhood consistency. Extensive experiments demonstrate that SG-GS achieves state-of-the-art geometry and appearance reconstruction performance.
Authors: Dongheon Lee, Seokju Yun, Youngmin Ro
Abstract: Recently, Super-Resolution (SR) achieved significant performance improvement by employing neural networks. Most SR methods conventionally train a single model for each targeted scale, which increases redundancy in training and deployment in proportion to the number of scales targeted. This paper challenges this conventional fixed-scale approach. Our preliminary analysis reveals that, surprisingly, encoders trained at different scales extract similar features from images. Furthermore, the commonly used scale-specific upsampler, Sub-Pixel Convolution (SPConv), exhibits significant inter-scale correlations. Based on these observations, we propose a framework for training multiple integer scales simultaneously with a single model. We use a single encoder to extract features and introduce a novel upsampler, Implicit Grid Convolution~(IGConv), which integrates SPConv at all scales within a single module to predict multiple scales. Our extensive experiments demonstrate that training multiple scales with a single model reduces the training budget and stored parameters by one-third while achieving equivalent inference latency and comparable performance. Furthermore, we propose IGConv$^{+}$, which addresses spectral bias and input-independent upsampling and uses ensemble prediction to improve performance. As a result, SRFormer-IGConv$^{+}$ achieves a remarkable 0.25dB improvement in PSNR at Urban100$\times$4 while reducing the training budget, stored parameters, and inference cost compared to the existing SRFormer.
Authors: Jingyao Wang, Luntian Mou, Changwen Zheng, Wen Gao
Abstract: Freeform handwriting authentication verifies a person's identity from their writing style and habits in messy handwriting data. This technique has gained widespread attention in recent years as a valuable tool for various fields, e.g., fraud prevention and cultural heritage protection. However, it still remains a challenging task in reality due to three reasons: (i) severe damage, (ii) complex high-dimensional features, and (iii) lack of supervision. To address these issues, we propose SherlockNet, an energy-oriented two-branch contrastive self-supervised learning framework for robust and fast freeform handwriting authentication. It consists of four stages: (i) pre-processing: converting manuscripts into energy distributions using a novel plug-and-play energy-oriented operator to eliminate the influence of noise; (ii) generalized pre-training: learning general representation through two-branch momentum-based adaptive contrastive learning with the energy distributions, which handles the high-dimensional features and spatial dependencies of handwriting; (iii) personalized fine-tuning: calibrating the learned knowledge using a small amount of labeled data from downstream tasks; and (iv) practical application: identifying individual handwriting from scrambled, missing, or forged data efficiently and conveniently. Considering the practicality, we construct EN-HA, a novel dataset that simulates data forgery and severe damage in real applications. Finally, we conduct extensive experiments on six benchmark datasets including our EN-HA, and the results prove the robustness and efficiency of SherlockNet.
Authors: Jialu Wang, Kaichen Zhou, Andrew Markham, Niki Trigoni
Abstract: Location information is pivotal for the automation and intelligence of terminal devices and edge-cloud IoT systems, such as autonomous vehicles and augmented reality. However, achieving reliable positioning across diverse IoT applications remains challenging due to significant training costs and the necessity of densely collected data. To tackle these issues, we have innovatively applied the selective state space (SSM) model to visual localization, introducing a new model named MambaLoc. The proposed model demonstrates exceptional training efficiency by capitalizing on the SSM model's strengths in efficient feature extraction, rapid computation, and memory optimization, and it further ensures robustness in sparse data environments due to its parameter sparsity. Additionally, we propose the Global Information Selector (GIS), which leverages selective SSM to implicitly achieve the efficient global feature extraction capabilities of Non-local Neural Networks. This design leverages the computational efficiency of the SSM model alongside the Non-local Neural Networks' capacity to capture long-range dependencies with minimal layers. Consequently, the GIS enables effective global information capture while significantly accelerating convergence. Our extensive experimental validation using public indoor and outdoor datasets first demonstrates our model's effectiveness, followed by evidence of its versatility with various existing localization models.
Authors: Ruofan Liang, Zan Gojcic, Merlin Nimier-David, David Acuna, Nandita Vijaykumar, Sanja Fidler, Zian Wang
Abstract: The correct insertion of virtual objects in images of real-world scenes requires a deep understanding of the scene's lighting, geometry and materials, as well as the image formation process. While recent large-scale diffusion models have shown strong generative and inpainting capabilities, we find that current models do not sufficiently "understand" the scene shown in a single picture to generate consistent lighting effects (shadows, bright reflections, etc.) while preserving the identity and details of the composited object. We propose using a personalized large diffusion model as guidance to a physically based inverse rendering process. Our method recovers scene lighting and tone-mapping parameters, allowing the photorealistic composition of arbitrary virtual objects in single frames or videos of indoor or outdoor scenes. Our physically based pipeline further enables automatic materials and tone-mapping refinement.
Authors: Xinyang Wang, Yi Yang, Minfeng Zhu, Kecheng Zheng, Shi Liu, Wei Chen
Abstract: Recent advancements in pre-trained Vision-Language Models (VLMs) have highlighted the significant potential of prompt tuning for adapting these models to a wide range of downstream tasks. However, existing prompt tuning methods typically map an image to a single representation, limiting the model's ability to capture the diverse ways an image can be described. To address this limitation, we investigate the impact of visual prompts on the model's generalization capability and introduce a novel method termed Multi-Representation Guided Prompt Tuning (MePT). Specifically, MePT employs a three-branch framework that focuses on diverse salient regions, uncovering the inherent knowledge within images which is crucial for robust generalization. Further, we employ efficient self-ensemble techniques to integrate these versatile image representations, allowing MePT to learn all conditional, marginal, and fine-grained distributions effectively. We validate the effectiveness of MePT through extensive experiments, demonstrating significant improvements on both base-to-novel class prediction and domain generalization tasks.
Authors: Cong Cong, Shiyu Xuan, Sidong Liu, Maurice Pagnucco, Shiliang Zhang, Yang Song
Abstract: Deep neural networks (DNNs) have exhibited remarkable success in the field of histopathology image analysis. On the other hand, the contemporary trend of employing large models and extensive datasets has underscored the significance of dataset distillation, which involves compressing large-scale datasets into a condensed set of synthetic samples, offering distinct advantages in improving training efficiency and streamlining downstream applications. In this work, we introduce a novel dataset distillation algorithm tailored for histopathology image datasets (Histo-DD), which integrates stain normalisation and model augmentation into the distillation progress. Such integration can substantially enhance the compatibility with histopathology images that are often characterised by high colour heterogeneity. We conduct a comprehensive evaluation of the effectiveness of the proposed algorithm and the generated histopathology samples in both patch-level and slide-level classification tasks. The experimental results, carried out on three publicly available WSI datasets, including Camelyon16, TCGA-IDH, and UniToPath, demonstrate that the proposed Histo-DD can generate more informative synthetic patches than previous coreset selection and patch sampling methods. Moreover, the synthetic samples can preserve discriminative information, substantially reduce training efforts, and exhibit architecture-agnostic properties. These advantages indicate that synthetic samples can serve as an alternative to large-scale datasets.
Authors: Jiandong Jin, Xiao Wang, Qian Zhu, Haiyang Wang, Chenglong Li
Abstract: Pedestrian Attribute Recognition (PAR) is one of the indispensable tasks in human-centered research. However, existing datasets neglect different domains (e.g., environments, times, populations, and data sources), only conducting simple random splits, and the performance of these datasets has already approached saturation. In the past five years, no large-scale dataset has been opened to the public. To address this issue, this paper proposes a new large-scale, cross-domain pedestrian attribute recognition dataset to fill the data gap, termed MSP60K. It consists of 60,122 images and 57 attribute annotations across eight scenarios. Synthetic degradation is also conducted to further narrow the gap between the dataset and real-world challenging scenarios. To establish a more rigorous benchmark, we evaluate 17 representative PAR models under both random and cross-domain split protocols on our dataset. Additionally, we propose an innovative Large Language Model (LLM) augmented PAR framework, named LLM-PAR. This framework processes pedestrian images through a Vision Transformer (ViT) backbone to extract features and introduces a multi-embedding query Transformer to learn partial-aware features for attribute classification. Significantly, we enhance this framework with LLM for ensemble learning and visual feature augmentation. Comprehensive experiments across multiple PAR benchmark datasets have thoroughly validated the efficacy of our proposed framework. The dataset and source code accompanying this paper will be made publicly available at \url{https://github.com/Event-AHU/OpenPAR}.
Authors: Yerim Jeon, Subeen Lee, Jihwan Kim, Jae-Pil Heo
Abstract: Few-shot object counting has garnered significant attention for its practicality as it aims to count target objects in a query image based on given exemplars without the need for additional training. However, there is a shortcoming in the prevailing extract-and-match approach: query and exemplar features lack interaction during feature extraction since they are extracted unaware of each other and later correlated based on similarity. This can lead to insufficient target awareness of the extracted features, resulting in target confusion in precisely identifying the actual target when multiple class objects coexist. To address this limitation, we propose a novel framework, Mutually-Aware FEAture learning(MAFEA), which encodes query and exemplar features mutually aware of each other from the outset. By encouraging interaction between query and exemplar features throughout the entire pipeline, we can obtain target-aware features that are robust to a multi-category scenario. Furthermore, we introduce a background token to effectively associate the target region of query with exemplars and decouple its background region from them. Our extensive experiments demonstrate that our model reaches a new state-of-the-art performance on the two challenging benchmarks, FSCD-LVIS and FSC-147, with a remarkably reduced degree of the target confusion problem.
Authors: Mingrui Wu, Oucheng Huang, Jiayi Ji, Jiale Li, Xinyue Cai, Huafeng Kuang, Jianzhuang Liu, Xiaoshuai Sun, Rongrong Ji
Abstract: In this work, we propose a training-free, trajectory-based controllable T2I approach, termed TraDiffusion. This novel method allows users to effortlessly guide image generation via mouse trajectories. To achieve precise control, we design a distance awareness energy function to effectively guide latent variables, ensuring that the focus of generation is within the areas defined by the trajectory. The energy function encompasses a control function to draw the generation closer to the specified trajectory and a movement function to diminish activity in areas distant from the trajectory. Through extensive experiments and qualitative assessments on the COCO dataset, the results reveal that TraDiffusion facilitates simpler, more natural image control. Moreover, it showcases the ability to manipulate salient regions, attributes, and relationships within the generated images, alongside visual input based on arbitrary or enhanced trajectories.
Authors: Xiao Wang, Yuehang Li, Fuling Wang, Shiao Wang, Chuanfu Li, Bo Jiang
Abstract: Inspired by the tremendous success of Large Language Models (LLMs), existing X-ray medical report generation methods attempt to leverage large models to achieve better performance. They usually adopt a Transformer to extract the visual features of a given X-ray image, and then, feed them into the LLM for text generation. How to extract more effective information for the LLMs to help them improve final results is an urgent problem that needs to be solved. Additionally, the use of visual Transformer models also brings high computational complexity. To address these issues, this paper proposes a novel context-guided efficient X-ray medical report generation framework. Specifically, we introduce the Mamba as the vision backbone with linear complexity, and the performance obtained is comparable to that of the strong Transformer model. More importantly, we perform context retrieval from the training set for samples within each mini-batch during the training phase, utilizing both positively and negatively related samples to enhance feature representation and discriminative learning. Subsequently, we feed the vision tokens, context information, and prompt statements to invoke the LLM for generating high-quality medical reports. Extensive experiments on three X-ray report generation datasets (i.e., IU-Xray, MIMIC-CXR, CheXpert Plus) fully validated the effectiveness of our proposed model. The source code of this work will be released on \url{https://github.com/Event-AHU/Medical_Image_Analysis}.
Authors: Zhendong Mao, Mengqi Huang, Fei Ding, Mingcong Liu, Qian He, Xiaojun Chang, Yongdong Zhang
Abstract: Text-to-image customization, which takes given texts and images depicting given subjects as inputs, aims to synthesize new images that align with both text semantics and subject appearance. This task provides precise control over details that text alone cannot capture and is fundamental for various real-world applications, garnering significant interest from academia and industry. Existing works follow the pseudo-word paradigm, which involves representing given subjects as pseudo-words and combining them with given texts to collectively guide the generation. However, the inherent conflict and entanglement between the pseudo-words and texts result in a dual-optimum paradox, where subject similarity and text controllability cannot be optimal simultaneously. We propose a novel real-words paradigm termed RealCustom++ that instead represents subjects as non-conflict real words, thereby disentangling subject similarity from text controllability and allowing both to be optimized simultaneously. Specifically, RealCustom++ introduces a novel "train-inference" decoupled framework: (1) During training, RealCustom++ learns the alignment between vision conditions and all real words in the text, ensuring high subject-similarity generation in open domains. This is achieved by the cross-layer cross-scale projector to robustly and finely extract subject features, and a curriculum training recipe that adapts the generated subject to diverse poses and sizes. (2) During inference, leveraging the learned general alignment, an adaptive mask guidance is proposed to only customize the generation of the specific target real word, keeping other subject-irrelevant regions uncontaminated to ensure high text-controllability in real-time.
Authors: Kun Luo, Bowen Zheng, Shidong Lv, Jie Tao, Qiang Wei
Abstract: Prostate cancer is the second most common cancer in males worldwide, and mpMRI is commonly used for diagnosis. However, interpreting mpMRI is challenging and requires expertise from radiologists. This highlights the urgent need for automated grading in mpMRI. Existing studies lack integration of clinical prior information and suffer from uneven training sample distribution due to prevalence. Therefore, we propose a solution that incorporates prior knowledge, addresses the issue of uneven medical sample distribution, and maintains high interpretability in mpMRI. Firstly, we introduce Prior Knowledge-Based Feature Extraction, which mathematically models the PI-RADS criteria for prostate cancer as diagnostic information into model training. Secondly, we propose Adaptive Recall Feedback Loss to address the extremely imbalanced data problem. This method adjusts the training dynamically based on accuracy and recall in the validation set, resulting in high accuracy and recall simultaneously in the testing set.Thirdly, we design an Enhanced Cascade Prostate Cancer Classifier that classifies prostate cancer into different levels in an interpretable way, which refines the classification results and helps with clinical intervention. Our method is validated through experiments on the PI-CAI dataset and outperforms other methods with a more balanced result in both accuracy and recall rate.
Authors: Hang Zou, Chenxi Du, Ajian Liu, Yuan Zhang, Jing Liu, Mingchuan Yang, Jun Wan, Hui Zhang
Abstract: Iris recognition is widely used in high-security scenarios due to its stability and distinctiveness. However, the acquisition of iris images typically requires near-infrared illumination and near-infrared band filters, leading to significant and consistent differences in imaging across devices. This underscores the importance of developing cross-domain capabilities in iris anti-spoofing methods. Despite this need, there is no dataset available that comprehensively evaluates the generalization ability of the iris anti-spoofing task. To address this gap, we propose the IrisGeneral dataset, which includes 10 subsets, belonging to 7 databases, published by 4 institutions, collected with 6 types of devices. IrisGeneral is designed with three protocols, aimed at evaluating average performance, cross-racial generalization, and cross-device generalization of iris anti-spoofing models. To tackle the challenge of integrating multiple sub-datasets in IrisGeneral, we employ multiple parameter sets to learn from the various subsets. Specifically, we utilize the Mixture of Experts (MoE) to fit complex data distributions using multiple sub-neural networks. To further enhance the generalization capabilities, we introduce a novel method Masked-MoE (MMoE). It randomly masks a portion of tokens for some experts and requires their outputs to be similar to the unmasked experts, which improves the generalization ability and effectively mitigates the overfitting issue produced by MoE. We selected ResNet50, VIT-B/16, CLIP, and FLIP as representative models and benchmarked them on the IrisGeneral dataset. Experimental results demonstrate that our proposed MMoE with CLIP achieves the best performance on IrisGeneral.
Authors: Xiao Wang, Shiao Wang, Pengpeng Shao, Bo Jiang, Lin Zhu, Yonghong Tian
Abstract: Human Action Recognition (HAR) stands as a pivotal research domain in both computer vision and artificial intelligence, with RGB cameras dominating as the preferred tool for investigation and innovation in this field. However, in real-world applications, RGB cameras encounter numerous challenges, including light conditions, fast motion, and privacy concerns. Consequently, bio-inspired event cameras have garnered increasing attention due to their advantages of low energy consumption, high dynamic range, etc. Nevertheless, most existing event-based HAR datasets are low resolution ($346 \times 260$). In this paper, we propose a large-scale, high-definition ($1280 \times 800$) human action recognition dataset based on the CeleX-V event camera, termed CeleX-HAR. It encompasses 150 commonly occurring action categories, comprising a total of 124,625 video sequences. Various factors such as multi-view, illumination, action speed, and occlusion are considered when recording these data. To build a more comprehensive benchmark dataset, we report over 20 mainstream HAR models for future works to compare. In addition, we also propose a novel Mamba vision backbone network for event stream based HAR, termed EVMamba, which equips the spatial plane multi-directional scanning and novel voxel temporal scanning mechanism. By encoding and mining the spatio-temporal information of event streams, our EVMamba has achieved favorable results across multiple datasets. Both the dataset and source code will be released on \url{https://github.com/Event-AHU/CeleX-HAR}
Authors: Yuxia Geng, Runkai Zhu, Jiaoyan Chen, Jintai Chen, Zhuo Chen, Xiang Chen, Can Xu, Yuxiang Wang, Xiaoliang Xu
Abstract: Disentanglement of visual features of primitives (i.e., attributes and objects) has shown exceptional results in Compositional Zero-shot Learning (CZSL). However, due to the feature divergence of an attribute (resp. object) when combined with different objects (resp. attributes), it is challenging to learn disentangled primitive features that are general across different compositions. To this end, we propose the solution of cross-composition feature disentanglement, which takes multiple primitive-sharing compositions as inputs and constrains the disentangled primitive features to be general across these compositions. More specifically, we leverage a compositional graph to define the overall primitive-sharing relationships between compositions, and build a task-specific architecture upon the recently successful large pre-trained vision-language model (VLM) CLIP, with dual cross-composition disentangling adapters (called L-Adapter and V-Adapter) inserted into CLIP's frozen text and image encoders, respectively. Evaluation on three popular CZSL benchmarks shows that our proposed solution significantly improves the performance of CZSL, and its components have been verified by solid ablation studies.
Authors: Syed Jawwad Haider Hamdani, Saifullah Saifullah, Stefan Agne, Andreas Dengel, Sheraz Ahmed
Abstract: Obtaining annotated table structure data for complex tables is a challenging task due to the inherent diversity and complexity of real-world document layouts. The scarcity of publicly available datasets with comprehensive annotations for intricate table structures hinders the development and evaluation of models designed for such scenarios. This research paper introduces a novel approach for generating annotated images for table structure by leveraging conditioned mask images of rows and columns through the application of latent diffusion models. The proposed method aims to enhance the quality of synthetic data used for training object detection models. Specifically, the study employs a conditioning mechanism to guide the generation of complex document table images, ensuring a realistic representation of table layouts. To evaluate the effectiveness of the generated data, we employ the popular YOLOv5 object detection model for training. The generated table images serve as valuable training samples, enriching the dataset with diverse table structures. The model is subsequently tested on the challenging pubtables-1m testset, a benchmark for table structure recognition in complex document layouts. Experimental results demonstrate that the introduced approach significantly improves the quality of synthetic data for training, leading to YOLOv5 models with enhanced performance. The mean Average Precision (mAP) values obtained on the pubtables-1m testset showcase results closely aligned with state-of-the-art methods. Furthermore, low FID results obtained on the synthetic data further validate the efficacy of the proposed methodology in generating annotated images for table structure.
Authors: Danush Kumar Venkatesh, Dominik Rivoir, Micha Pfeiffer, Stefanie Speidel
Abstract: Computer-assisted surgery (CAS) systems are designed to assist surgeons during procedures, thereby reducing complications and enhancing patient care. Training machine learning models for these systems requires a large corpus of annotated datasets, which is challenging to obtain in the surgical domain due to patient privacy concerns and the significant labeling effort required from doctors. Previous methods have explored unpaired image translation using generative models to create realistic surgical images from simulations. However, these approaches have struggled to produce high-quality, diverse surgical images. In this work, we introduce \emph{SurgicaL-CD}, a consistency-distilled diffusion method to generate realistic surgical images with only a few sampling steps without paired data. We evaluate our approach on three datasets, assessing the generated images in terms of quality and utility as downstream training datasets. Our results demonstrate that our method outperforms GANs and diffusion-based approaches. Our code is available at \url{https://gitlab.com/nct_tso_public/gan2diffusion}.
Authors: Jun Yan, Pengyu Wang, Danni Wang, Weiquan Huang, Daniel Watzenig, Huilin Yin
Abstract: Semantic segmentation is a significant perception task in autonomous driving. It suffers from the risks of adversarial examples. In the past few years, deep learning has gradually transitioned from convolutional neural network (CNN) models with a relatively small number of parameters to foundation models with a huge number of parameters. The segment-anything model (SAM) is a generalized image segmentation framework that is capable of handling various types of images and is able to recognize and segment arbitrary objects in an image without the need to train on a specific object. It is a unified model that can handle diverse downstream tasks, including semantic segmentation, object detection, and tracking. In the task of semantic segmentation for autonomous driving, it is significant to study the zero-shot adversarial robustness of SAM. Therefore, we deliver a systematic empirical study on the robustness of SAM without additional training. Based on the experimental results, the zero-shot adversarial robustness of the SAM under the black-box corruptions and white-box adversarial attacks is acceptable, even without the need for additional training. The finding of this study is insightful in that the gigantic model parameters and huge amounts of training data lead to the phenomenon of emergence, which builds a guarantee of adversarial robustness. SAM is a vision foundation model that can be regarded as an early prototype of an artificial general intelligence (AGI) pipeline. In such a pipeline, a unified model can handle diverse tasks. Therefore, this research not only inspects the impact of vision foundation models on safe autonomous driving but also provides a perspective on developing trustworthy AGI. The code is available at: https://github.com/momo1986/robust_sam_iv.
Authors: Heng Li, Yuenan Hou, Xiaohan Xing, Xiao Sun, Yanyong Zhang
Abstract: Training deep learning models for semantic occupancy prediction is challenging due to factors such as a large number of occupancy cells, severe occlusion, limited visual cues, complicated driving scenarios, etc. Recent methods often adopt transformer-based architectures given their strong capability in learning input-conditioned weights and long-range relationships. However, transformer-based networks are notorious for their quadratic computation complexity, seriously undermining their efficacy and deployment in semantic occupancy prediction. Inspired by the global modeling and linear computation complexity of the Mamba architecture, we present the first Mamba-based network for semantic occupancy prediction, termed OccMamba. However, directly applying the Mamba architecture to the occupancy prediction task yields unsatisfactory performance due to the inherent domain gap between the linguistic and 3D domains. To relieve this problem, we present a simple yet effective 3D-to-1D reordering operation, i.e., height-prioritized 2D Hilbert expansion. It can maximally retain the spatial structure of point clouds as well as facilitate the processing of Mamba blocks. Our OccMamba achieves state-of-the-art performance on three prevalent occupancy prediction benchmarks, including OpenOccupancy, SemanticKITTI and SemanticPOSS. Notably, on OpenOccupancy, our OccMamba outperforms the previous state-of-the-art Co-Occ by 3.1% IoU and 3.2% mIoU, respectively. Codes will be released upon publication.
Authors: Yash Bhalgat, Vadim Tschernezki, Iro Laina, Jo\~ao F. Henriques, Andrea Vedaldi, Andrew Zisserman
Abstract: Egocentric videos present unique challenges for 3D scene understanding due to rapid camera motion, frequent object occlusions, and limited object visibility. This paper introduces a novel approach to instance segmentation and tracking in first-person video that leverages 3D awareness to overcome these obstacles. Our method integrates scene geometry, 3D object centroid tracking, and instance segmentation to create a robust framework for analyzing dynamic egocentric scenes. By incorporating spatial and temporal cues, we achieve superior performance compared to state-of-the-art 2D approaches. Extensive evaluations on the challenging EPIC Fields dataset demonstrate significant improvements across a range of tracking and segmentation consistency metrics. Specifically, our method outperforms the next best performing approach by $7$ points in Association Accuracy (AssA) and $4.5$ points in IDF1 score, while reducing the number of ID switches by $73\%$ to $80\%$ across various object categories. Leveraging our tracked instance segmentations, we showcase downstream applications in 3D object reconstruction and amodal video object segmentation in these egocentric settings.
Authors: Sihan Yang, Haixia Bi, Hai Zhang, Jian Sun
Abstract: Segment Anything Model (SAM) has demonstrated impressive performance on a wide range of natural image segmentation tasks. However, its performance significantly deteriorates when directly applied to medical domain, due to the remarkable differences between natural images and medical images. Some researchers have attempted to train SAM on large scale medical datasets. However, poor zero-shot performance is observed from the experimental results. In this context, inspired by the superior performance of U-Net-like models in medical image segmentation, we propose SAMUNet, a new foundation model which incorporates U-Net to the original SAM, to fully leverage the powerful contextual modeling ability of convolutions. To be specific, we parallel a convolutional branch in the image encoder, which is trained independently with the vision Transformer branch frozen. Additionally, we employ multi-scale fusion in the mask decoder, to facilitate accurate segmentation of objects with different scales. We train SAM-UNet on SA-Med2D-16M, the largest 2-dimensional medical image segmentation dataset to date, yielding a universal pretrained model for medical images. Extensive experiments are conducted to evaluate the performance of the model, and state-of-the-art result is achieved, with a dice similarity coefficient score of 0.883 on SA-Med2D-16M dataset. Specifically, in zero-shot segmentation experiments, our model not only significantly outperforms previous large medical SAM models across all modalities, but also substantially mitigates the performance degradation seen on unseen modalities. It should be highlighted that SAM-UNet is an efficient and extensible foundation model, which can be further fine-tuned for other downstream tasks in medical community. The code is available at https://github.com/Hhankyangg/sam-unet.
Authors: Alik Pramanick, Arijit Sur, V. Vijaya Saradhi
Abstract: Underwater imagery is often compromised by factors such as color distortion and low contrast, posing challenges for high-level vision tasks. Recent underwater image restoration (UIR) methods either analyze the input image at full resolution, resulting in spatial richness but contextual weakness, or progressively from high to low resolution, yielding reliable semantic information but reduced spatial accuracy. Here, we propose a lightweight multi-stage network called Lit-Net that focuses on multi-resolution and multi-scale image analysis for restoring underwater images while retaining original resolution during the first stage, refining features in the second, and focusing on reconstruction in the final stage. Our novel encoder block utilizes parallel $1\times1$ convolution layers to capture local information and speed up operations. Further, we incorporate a modified weighted color channel-specific $l_1$ loss ($cl_1$) function to recover color and detail information. Extensive experimentations on publicly available datasets suggest our model's superiority over recent state-of-the-art methods, with significant improvement in qualitative and quantitative measures, such as $29.477$ dB PSNR ($1.92\%$ improvement) and $0.851$ SSIM ($2.87\%$ improvement) on the EUVP dataset. The contributions of Lit-Net offer a more robust approach to underwater image enhancement and super-resolution, which is of considerable importance for underwater autonomous vehicles and surveillance. The code is available at: https://github.com/Alik033/Lit-Net.
Authors: Qizhou Chen, Taolin Zhang, Chengyu Wang, Xiaofeng He, Dakan Wang, Tingting Liu
Abstract: Model editing aims to correct outdated or erroneous knowledge in large models without costly retraining. Recent research discovered that the mid-layer representation of the subject's final token in a prompt has a strong influence on factual predictions, and developed Large Language Model (LLM) editing techniques based on this observation. However, for Vision-LLMs (VLLMs), how visual representations impact the predictions from a decoder-only language model remains largely unexplored. To the best of our knowledge, model editing for VLLMs has not been extensively studied in the literature. In this work, we employ the contribution allocation and noise perturbation methods to measure the contributions of visual representations for token predictions. Our attribution analysis shows that visual representations in mid-to-later layers that are highly relevant to the prompt contribute significantly to predictions. Based on these insights, we propose VisEdit, a novel model editor for VLLMs that effectively corrects knowledge by editing intermediate visual representations in regions important to the edit prompt. We evaluated VisEdit using multiple VLLM backbones and public VLLM editing benchmark datasets. The results show the superiority of VisEdit over the strong baselines adapted from existing state-of-the-art editors for LLMs.
Authors: Zhanzhong Pang, Fadime Sener, Shrinivas Ramasubramanian, Angela Yao
Abstract: Procedural activity videos often exhibit a long-tailed action distribution due to varying action frequencies and durations. However, state-of-the-art temporal action segmentation methods overlook the long tail and fail to recognize tail actions. Existing long-tail methods make class-independent assumptions and struggle to identify tail classes when applied to temporal segmentation frameworks. This work proposes a novel group-wise temporal logit adjustment~(G-TLA) framework that combines a group-wise softmax formulation while leveraging activity information and action ordering for logit adjustment. The proposed framework significantly improves in segmenting tail actions without any performance loss on head actions.
Authors: Kang Xiao, Xu Wang, Yulin He, Baoliang Chen, Xuelin Shen
Abstract: Full-reference image quality assessment (FR-IQA) models generally operate by measuring the visual differences between a degraded image and its reference. However, existing FR-IQA models including both the classical ones (eg, PSNR and SSIM) and deep-learning based measures (eg, LPIPS and DISTS) still exhibit limitations in capturing the full perception characteristics of the human visual system (HVS). In this paper, instead of designing a new FR-IQA measure, we aim to explore a generalized human visual attention estimation strategy to mimic the process of human quality rating and enhance existing IQA models. In particular, we model human attention generation by measuring the statistical dependency between the degraded image and the reference image. The dependency is captured in a training-free manner by our proposed sliced maximal information coefficient and exhibits surprising generalization in different IQA measures. Experimental results verify the performance of existing IQA models can be consistently improved when our attention module is incorporated. The source code is available at https://github.com/KANGX99/SMIC.
Authors: Corentin Dumery, Aoxiang Fan, Ren Li, Nicolas Talabot, Pascal Fua
Abstract: Neural Radiance Fields (NeRFs) have become a powerful tool for modeling 3D scenes from multiple images. However, NeRFs remain difficult to segment into semantically meaningful regions. Previous approaches to 3D segmentation of NeRFs either require user interaction to isolate a single object, or they rely on 2D semantic masks with a limited number of classes for supervision. As a consequence, they generalize poorly to class-agnostic masks automatically generated in real scenes. This is attributable to the ambiguity arising from zero-shot segmentation, yielding inconsistent masks across views. In contrast, we propose a method that is robust to inconsistent segmentations and successfully decomposes the scene into a set of objects of any class. By introducing a limited number of competing object slots against which masks are matched, a meaningful object representation emerges that best explains the 2D supervision and minimizes an additional regularization term. Our experiments demonstrate the ability of our method to generate 3D panoptic segmentations on complex scenes, and extract high-quality 3D assets from NeRFs that can then be used in virtual 3D environments.
Authors: Alik Pramanick, Utsav Bheda, Arijit Sur
Abstract: Recently, transformers have captured significant interest in the area of single-image super-resolution tasks, demonstrating substantial gains in performance. Current models heavily depend on the network's extensive ability to extract high-level semantic details from images while overlooking the effective utilization of multi-scale image details and intermediate information within the network. Furthermore, it has been observed that high-frequency areas in images present significant complexity for super-resolution compared to low-frequency areas. This work proposes a transformer-based super-resolution architecture called ML-CrAIST that addresses this gap by utilizing low-high frequency information in multiple scales. Unlike most of the previous work (either spatial or channel), we operate spatial and channel self-attention, which concurrently model pixel interaction from both spatial and channel dimensions, exploiting the inherent correlations across spatial and channel axis. Further, we devise a cross-attention block for super-resolution, which explores the correlations between low and high-frequency information. Quantitative and qualitative assessments indicate that our proposed ML-CrAIST surpasses state-of-the-art super-resolution methods (e.g., 0.15 dB gain @Manga109 $\times$4). Code is available on: https://github.com/Alik033/ML-CrAIST.
Authors: Dario Zanca, Andrea Zugarini, Simon Dietz, Thomas R. Altstidl, Mark A. Turban Ndjeuha, Leo Schwinn, Bjoern Eskofier
Abstract: Understanding human attention is crucial for vision science and AI. While many models exist for free-viewing, less is known about task-driven image exploration. To address this, we introduce CapMIT1003, a dataset with captions and click-contingent image explorations, to study human attention during the captioning task. We also present NevaClip, a zero-shot method for predicting visual scanpaths by combining CLIP models with NeVA algorithms. NevaClip generates fixations to align the representations of foveated visual stimuli and captions. The simulated scanpaths outperform existing human attention models in plausibility for captioning and free-viewing tasks. This research enhances the understanding of human attention and advances scanpath prediction models.
Authors: Zhigang Chen, Benjia Zhou, Yiqing Huang, Jun Wan, Yibo Hu, Hailin Shi, Yanyan Liang, Zhen Lei, Du Zhang
Abstract: Sign Language Representation Learning (SLRL) is crucial for a range of sign language-related downstream tasks such as Sign Language Translation (SLT) and Sign Language Retrieval (SLRet). Recently, many gloss-based and gloss-free SLRL methods have been proposed, showing promising performance. Among them, the gloss-free approach shows promise for strong scalability without relying on gloss annotations. However, it currently faces suboptimal solutions due to challenges in encoding the intricate, context-sensitive characteristics of sign language videos, mainly struggling to discern essential sign features using a non-monotonic video-text alignment strategy. Therefore, we introduce an innovative pretraining paradigm for gloss-free SLRL, called C${^2}$RL, in this paper. Specifically, rather than merely incorporating a non-monotonic semantic alignment of video and text to learn language-oriented sign features, we emphasize two pivotal aspects of SLRL: Implicit Content Learning (ICL) and Explicit Context Learning (ECL). ICL delves into the content of communication, capturing the nuances, emphasis, timing, and rhythm of the signs. In contrast, ECL focuses on understanding the contextual meaning of signs and converting them into equivalent sentences. Despite its simplicity, extensive experiments confirm that the joint optimization of ICL and ECL results in robust sign language representation and significant performance gains in gloss-free SLT and SLRet tasks. Notably, C${^2}$RL improves the BLEU-4 score by +5.3 on P14T, +10.6 on CSL-daily, +6.2 on OpenASL, and +1.3 on How2Sign. It also boosts the R@1 score by +8.3 on P14T, +14.4 on CSL-daily, and +5.9 on How2Sign. Additionally, we set a new baseline for the OpenASL dataset in the SLRet task.
Authors: Ik Jun Moon, Junho Moon, Ikbeom Jang
Abstract: 1. Research question: With the growing interest in skin diseases and skin aesthetics, the ability to predict facial wrinkles is becoming increasingly important. This study aims to evaluate whether a computational model, convolutional neural networks (CNN), can be trained for automated facial wrinkle segmentation. 2. Findings: Our study presents an effective technique for integrating data from multiple annotators and illustrates that transfer learning can enhance performance, resulting in dependable segmentation of facial wrinkles. 3. Meaning: This approach automates intricate and time-consuming tasks of wrinkle analysis with a deep learning framework. It could be used to facilitate skin treatments and diagnostics.
Authors: Yadong Lu, Shitian Zhao, Boxiang Yun, Dongsheng Jiang, Yin Li, Qingli Li, Yan Wang
Abstract: Despite recent progress in enhancing the efficacy of Open-Domain Continual Learning (ODCL) in Vision-Language Models (VLM), failing to (1) correctly identify the Task-ID of a test image and (2) use only the category set corresponding to the Task-ID, while preserving the knowledge related to each domain, cannot address the two primary challenges of ODCL: forgetting old knowledge and maintaining zero-shot capabilities, as well as the confusions caused by category-relatedness between domains. In this paper, we propose a simple yet effective solution: leveraging intra-domain category-aware prototypes for ODCL in CLIP (DPeCLIP), where the prototype is the key to bridging the above two processes. Concretely, we propose a training-free Task-ID discriminator method, by utilizing prototypes as classifiers for identifying Task-IDs. Furthermore, to maintain the knowledge corresponding to each domain, we incorporate intra-domain category-aware prototypes as domain prior prompts into the training process. Extensive experiments conducted on 11 different datasets demonstrate the effectiveness of our approach, achieving 2.37% and 1.14% average improvement in class-incremental and task-incremental settings, respectively.
Authors: Xuechao Chen, Ying Chen, Jialin Li, Qiang Nie, Yong Liu, Qixing Huang, Yang Li
Abstract: 3D pre-training is crucial to 3D perception tasks. However, limited by the difficulties in collecting clean 3D data, 3D pre-training consistently faced data scaling challenges. Inspired by semi-supervised learning leveraging limited labeled data and a large amount of unlabeled data, in this work, we propose a novel self-supervised pre-training framework utilizing the real 3D data and the pseudo-3D data lifted from images by a large depth estimation model. Another challenge lies in the efficiency. Previous methods such as Point-BERT and Point-MAE, employ k nearest neighbors to embed 3D tokens, requiring quadratic time complexity. To efficiently pre-train on such a large amount of data, we propose a linear-time-complexity token embedding strategy and a training-efficient 2D reconstruction target. Our method achieves state-of-the-art performance in 3D classification and few-shot learning while maintaining high pre-training and downstream fine-tuning efficiency.
Authors: Chen Feng, Georgios Tzimiropoulos, Ioannis Patras
Abstract: Learning with Noisy labels (LNL) poses a significant challenge for the Machine Learning community. Some of the most widely used approaches that select as clean samples for which the model itself (the in-training model) has high confidence, e.g., `small loss', can suffer from the so called `self-confirmation' bias. This bias arises because the in-training model, is at least partially trained on the noisy labels. Furthermore, in the classification case, an additional challenge arises because some of the label noise is between classes that are visually very similar (`hard noise'). This paper addresses these challenges by proposing a method (\textit{CLIPCleaner}) that leverages CLIP, a powerful Vision-Language (VL) model for constructing a zero-shot classifier for efficient, offline, clean sample selection. This has the advantage that the sample selection is decoupled from the in-training model and that the sample selection is aware of the semantic and visual similarities between the classes due to the way that CLIP is trained. We provide theoretical justifications and empirical evidence to demonstrate the advantages of CLIP for LNL compared to conventional pre-trained models. Compared to current methods that combine iterative sample selection with various techniques, \textit{CLIPCleaner} offers a simple, single-step approach that achieves competitive or superior performance on benchmark datasets. To the best of our knowledge, this is the first time a VL model has been used for sample selection to address the problem of Learning with Noisy Labels (LNL), highlighting their potential in the domain.
Authors: Kira Maag, Roman Resner, Asja Fischer
Abstract: Deep neural networks have demonstrated remarkable effectiveness across a wide range of tasks such as semantic segmentation. Nevertheless, these networks are vulnerable to adversarial attacks that add imperceptible perturbations to the input image, leading to false predictions. This vulnerability is particularly dangerous in safety-critical applications like automated driving. While adversarial examples and defense strategies are well-researched in the context of image classification, there is comparatively less research focused on semantic segmentation. Recently, we have proposed an uncertainty-based method for detecting adversarial attacks on neural networks for semantic segmentation. We observed that uncertainty, as measured by the entropy of the output distribution, behaves differently on clean versus adversely perturbed images, and we utilize this property to differentiate between the two. In this extended version of our work, we conduct a detailed analysis of uncertainty-based detection of adversarial attacks including a diverse set of adversarial attacks and various state-of-the-art neural networks. Our numerical experiments show the effectiveness of the proposed uncertainty-based detection method, which is lightweight and operates as a post-processing step, i.e., no model modifications or knowledge of the adversarial example generation process are required.
Authors: Vinit Hegiste, Tatjana Legler, Martin Ruskowski
Abstract: In the realm of Federated Learning (FL), particularly within the manufacturing sector, the strategy for selecting client weights for server aggregation is pivotal for model performance. This study investigates the comparative effectiveness of two weight selection strategies: Final Epoch Weight Selection (FEWS) and Optimal Epoch Weight Selection (OEWS). Designed for manufacturing contexts where collaboration typically involves a limited number of partners (two to four clients), our research focuses on federated image classification tasks. We employ various neural network architectures, including EfficientNet, ResNet, and VGG, to assess the impact of these weight selection strategies on model convergence and robustness. Our research aims to determine whether FEWS or OEWS enhances the global FL model's performance across communication rounds (CRs). Through empirical analysis and rigorous experimentation, we seek to provide valuable insights for optimizing FL implementations in manufacturing, ensuring that collaborative efforts yield the most effective and reliable models with a limited number of participating clients. The findings from this study are expected to refine FL practices significantly in manufacturing, thereby enhancing the efficiency and performance of collaborative machine learning endeavors in this vital sector.
Authors: Emanuele Caruso, Francesco Pelosin, Alessandro Simoni, Marco Boschetti
Abstract: In this work, we propose a simple yet effective method to tackle the problem of imbalanced multi-class semantic segmentation in deep learning systems. One of the key properties for a good training set is the balancing among the classes. When the input distribution is heavily imbalanced in the number of instances, the learning process could be hindered or difficult to carry on. To this end, we propose a Dynamic Label Injection (DLI) algorithm to impose a uniform distribution in the input batch. Our algorithm computes the current batch defect distribution and re-balances it by transferring defects using a combination of Poisson-based seamless image cloning and cut-paste techniques. A thorough experimental section on the Magnetic Tiles dataset shows better results of DLI compared to other balancing loss approaches also in the challenging weakly-supervised setup. The code is available at https://github.com/covisionlab/dynamic-label-injection.git
URLs: https://github.com/covisionlab/dynamic-label-injection.git
Authors: Wiktor Mucha, Michael Wray, Martin Kampel
Abstract: Hand pose represents key information for action recognition in the egocentric perspective, where the user is interacting with objects. We propose to improve egocentric 3D hand pose estimation based on RGB frames only by using pseudo-depth images. Incorporating state-of-the-art single RGB image depth estimation techniques, we generate pseudo-depth representations of the frames and use distance knowledge to segment irrelevant parts of the scene. The resulting depth maps are then used as segmentation masks for the RGB frames. Experimental results on H2O Dataset confirm the high accuracy of the estimated pose with our method in an action recognition task. The 3D hand pose, together with information from object detection, is processed by a transformer-based action recognition network, resulting in an accuracy of 91.73%, outperforming all state-of-the-art methods. Estimations of 3D hand pose result in competitive performance with existing methods with a mean pose error of 28.66 mm. This method opens up new possibilities for employing distance information in egocentric 3D hand pose estimation without relying on depth sensors.
Authors: Minye Wu, Tinne Tuytelaars
Abstract: Recent advancements in photo-realistic novel view synthesis have been significantly driven by Gaussian Splatting (3DGS). Nevertheless, the explicit nature of 3DGS data entails considerable storage requirements, highlighting a pressing need for more efficient data representations. To address this, we present Implicit Gaussian Splatting (IGS), an innovative hybrid model that integrates explicit point clouds with implicit feature embeddings through a multi-level tri-plane architecture. This architecture features 2D feature grids at various resolutions across different levels, facilitating continuous spatial domain representation and enhancing spatial correlations among Gaussian primitives. Building upon this foundation, we introduce a level-based progressive training scheme, which incorporates explicit spatial regularization. This method capitalizes on spatial correlations to enhance both the rendering quality and the compactness of the IGS representation. Furthermore, we propose a novel compression pipeline tailored for both point clouds and 2D feature grids, considering the entropy variations across different levels. Extensive experimental evaluations demonstrate that our algorithm can deliver high-quality rendering using only a few MBs, effectively balancing storage efficiency and rendering fidelity, and yielding results that are competitive with the state-of-the-art.
Authors: Junho Moon, Haejun Chung, Ikbeom Jang
Abstract: Facial wrinkle detection plays a crucial role in cosmetic dermatology. Precise manual segmentation of facial wrinkles is challenging and time-consuming, with inherent subjectivity leading to inconsistent results among graders. To address this issue, we propose two solutions. First, we build and release the first public facial wrinkle dataset, `FFHQ-Wrinkle', an extension of the NVIDIA FFHQ dataset. This dataset includes 1,000 images with human labels and 50,000 images with automatically generated weak labels. This dataset can foster the research community to develop advanced wrinkle detection algorithms. Second, we introduce a training strategy for U-Net-like encoder-decoder models to detect wrinkles across the face automatically. Our method employs a two-stage training strategy: texture map pretraining and finetuning on human-labeled data. Initially, we pretrain models on a large dataset with weak labels (N=50k) or masked texture maps generated through computer vision techniques, without human intervention. Subsequently, we finetune the models using human-labeled data (N=1k), which consists of manually labeled wrinkle masks. During finetuning, the network inputs a combination of RGB and masked texture maps, comprising four channels. We effectively combine labels from multiple annotators to minimize subjectivity in manual labeling. Our strategies demonstrate improved segmentation performance in facial wrinkle segmentation both quantitatively and visually compared to existing pretraining methods.
Authors: Reuben Dorent, Roya Khajavi, Tagwa Idris, Erik Ziegler, Bhanusupriya Somarouthu, Heather Jacene, Ann LaCasce, Jonathan Deissler, Jan Ehrhardt, Sofija Engelson, Stefan M. Fischer, Yun Gu, Heinz Handels, Satoshi Kasai, Satoshi Kondo, Klaus Maier-Hein, Julia A. Schnabel, Guotai Wang, Litingyu Wang, Tassilo Wald, Guang-Zhong Yang, Hanxiao Zhang, Minghui Zhang, Steve Pieper, Gordon Harris, Ron Kikinis, Tina Kapur
Abstract: Accurate assessment of lymph node size in 3D CT scans is crucial for cancer staging, therapeutic management, and monitoring treatment response. Existing state-of-the-art segmentation frameworks in medical imaging often rely on fully annotated datasets. However, for lymph node segmentation, these datasets are typically small due to the extensive time and expertise required to annotate the numerous lymph nodes in 3D CT scans. Weakly-supervised learning, which leverages incomplete or noisy annotations, has recently gained interest in the medical imaging community as a potential solution. Despite the variety of weakly-supervised techniques proposed, most have been validated only on private datasets or small publicly available datasets. To address this limitation, the Mediastinal Lymph Node Quantification (LNQ) challenge was organized in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2023). This challenge aimed to advance weakly-supervised segmentation methods by providing a new, partially annotated dataset and a robust evaluation framework. A total of 16 teams from 5 countries submitted predictions to the validation leaderboard, and 6 teams from 3 countries participated in the evaluation phase. The results highlighted both the potential and the current limitations of weakly-supervised approaches. On one hand, weakly-supervised approaches obtained relatively good performance with a median Dice score of $61.0\%$. On the other hand, top-ranked teams, with a median Dice score exceeding $70\%$, boosted their performance by leveraging smaller but fully annotated datasets to combine weak supervision and full supervision. This highlights both the promise of weakly-supervised methods and the ongoing need for high-quality, fully annotated data to achieve higher segmentation performance.
Authors: Zhengchao Huang, Bin Xia, Zicheng Lin, Zhun Mou, Wenming Yang
Abstract: The rapid advancement of deepfake technologies has sparked widespread public concern, particularly as face forgery poses a serious threat to public information security. However, the unknown and diverse forgery techniques, varied facial features and complex environmental factors pose significant challenges for face forgery analysis. Existing datasets lack descriptions of these aspects, making it difficult for models to distinguish between real and forged faces using only visual information amid various confounding factors. In addition, existing methods do not yield user-friendly and explainable results, complicating the understanding of the model's decision-making process. To address these challenges, we introduce a novel Open-World Face Forgery Analysis VQA (OW-FFA-VQA) task and the corresponding benchmark. To tackle this task, we first establish a dataset featuring a diverse collection of real and forged face images with essential descriptions and reliable forgery reasoning. Base on this dataset, we introduce FFAA: Face Forgery Analysis Assistant, consisting of a fine-tuned Multimodal Large Language Model (MLLM) and Multi-answer Intelligent Decision System (MIDS). By integrating hypothetical prompts with MIDS, the impact of fuzzy classification boundaries is effectively mitigated, enhancing the model's robustness. Extensive experiments demonstrate that our method not only provides user-friendly explainable results but also significantly boosts accuracy and robustness compared to previous methods.
Authors: Oliver Cory, Ozge Mercanoglu Sincan, Matthew Vowels, Alessia Battisti, Franz Holzknecht, Katja Tissi, Sandra Sidler-Miserez, Tobias Haug, Sarah Ebling, Richard Bowden
Abstract: Sign Language Assessment (SLA) tools are useful to aid in language learning and are underdeveloped. Previous work has focused on isolated signs or comparison against a single reference video to assess Sign Languages (SL). This paper introduces a novel SLA tool designed to evaluate the comprehensibility of SL by modelling the natural distribution of human motion. We train our pipeline on data from native signers and evaluate it using SL learners. We compare our results to ratings from a human raters study and find strong correlation between human ratings and our tool. We visually demonstrate our tools ability to detect anomalous results spatio-temporally, providing actionable feedback to aid in SL learning and assessment.
Authors: Tao Yang, Yangming Shi, Yunwen Huang, Feng Chen, Yin Zheng, Lei Zhang
Abstract: Text-to-video (T2V) generation has gained significant attention due to its wide applications to video generation, editing, enhancement and translation, \etc. However, high-quality (HQ) video synthesis is extremely challenging because of the diverse and complex motions existed in real world. Most existing works struggle to address this problem by collecting large-scale HQ videos, which are inaccessible to the community. In this work, we show that publicly available limited and low-quality (LQ) data are sufficient to train a HQ video generator without recaptioning or finetuning. We factorize the whole T2V generation process into two steps: generating an image conditioned on a highly descriptive caption, and synthesizing the video conditioned on the generated image and a concise caption of motion details. Specifically, we present \emph{Factorized-Dreamer}, a factorized spatiotemporal framework with several critical designs for T2V generation, including an adapter to combine text and image embeddings, a pixel-aware cross attention module to capture pixel-level image information, a T5 text encoder to better understand motion description, and a PredictNet to supervise optical flows. We further present a noise schedule, which plays a key role in ensuring the quality and stability of video generation. Our model lowers the requirements in detailed captions and HQ videos, and can be directly trained on limited LQ datasets with noisy and brief captions such as WebVid-10M, largely alleviating the cost to collect large-scale HQ video-text pairs. Extensive experiments in a variety of T2V and image-to-video generation tasks demonstrate the effectiveness of our proposed Factorized-Dreamer. Our source codes are available at \url{https://github.com/yangxy/Factorized-Dreamer/}.
Authors: Feiyu Pan, Hao Fang, Runmin Cong, Wei Zhang, Xiankai Lu
Abstract: Video Object Segmentation (VOS) task aims to segmenting a particular object instance throughout the entire video sequence given only the object mask of the first frame. Recently, Segment Anything Model 2 (SAM 2) is proposed, which is a foundation model towards solving promptable visual segmentation in images and videos. SAM 2 builds a data engine, which improves model and data via user interaction, to collect the largest video segmentation dataset to date. SAM 2 is a simple transformer architecture with streaming memory for real-time video processing, which trained on the date provides strong performance across a wide range of tasks. In this work, we evaluate the zero-shot performance of SAM 2 on the more challenging VOS datasets MOSE and LVOS. Without fine-tuning on the training set, SAM 2 achieved 75.79 J&F on the test set and ranked 4th place for 6th LSVOS Challenge VOS Track.
Authors: Hao Fang, Feiyu Pan, Xiankai Lu, Wei Zhang, Runmin Cong
Abstract: Referring video object segmentation (RVOS) relies on natural language expressions to segment target objects in video. In this year, LSVOS Challenge RVOS Track replaced the origin YouTube-RVOS benchmark with MeViS. MeViS focuses on referring the target object in a video through its motion descriptions instead of static attributes, posing a greater challenge to RVOS task. In this work, we integrate strengths of that leading RVOS and VOS models to build up a simple and effective pipeline for RVOS. Firstly, We finetune the state-of-the-art RVOS model to obtain mask sequences that are correlated with language descriptions. Secondly, based on a reliable and high-quality key frames, we leverage VOS model to enhance the quality and temporal consistency of the mask results. Finally, we further improve the performance of the RVOS model using semi-supervised learning. Our solution achieved 62.57 J&F on the MeViS test set and ranked 1st place for 6th LSVOS Challenge RVOS Track.
Authors: Wei Zhou, Zhou Wang
Abstract: Depth perception plays an essential role in the viewer experience for immersive virtual reality (VR) visual environments. However, previous research investigations in the depth quality of 3D/stereoscopic images are rather limited, and in particular, are largely lacking for 3D viewing of 360-degree omnidirectional content. In this work, we make one of the first attempts to develop an objective quality assessment model named depth quality index (DQI) for efficient no-reference (NR) depth quality assessment of stereoscopic omnidirectional images. Motivated by the perceptual characteristics of the human visual system (HVS), the proposed DQI is built upon multi-color-channel, adaptive viewport selection, and interocular discrepancy features. Experimental results demonstrate that the proposed method outperforms state-of-the-art image quality assessment (IQA) and depth quality assessment (DQA) approaches in predicting the perceptual depth quality when tested using both single-viewport and omnidirectional stereoscopic image databases. Furthermore, we demonstrate that combining the proposed depth quality model with existing IQA methods significantly boosts the performance in predicting the overall quality of 3D omnidirectional images.
Authors: Haoyang Wang, Liming Liu, Quanlu Jia, Jiangkai Wu, Haodan Zhang, Peiheng Wang, Xinggong Zhang
Abstract: Mesh reconstruction based on Neural Radiance Fields (NeRF) is popular in a variety of applications such as computer graphics, virtual reality, and medical imaging due to its efficiency in handling complex geometric structures and facilitating real-time rendering. However, existing works often fail to capture fine geometric details accurately and struggle with optimizing rendering quality. To address these challenges, we propose a novel algorithm that progressively generates and optimizes meshes from multi-view images. Our approach initiates with the training of a NeRF model to establish an initial Signed Distance Field (SDF) and a view-dependent appearance field. Subsequently, we iteratively refine the SDF through a differentiable mesh extraction method, continuously updating both the vertex positions and their connectivity based on the loss from mesh differentiable rasterization, while also optimizing the appearance representation. To further leverage high-fidelity and detail-rich representations from NeRF, we propose an online-learning strategy based on Upper Confidence Bound (UCB) to enhance viewpoints by adaptively incorporating images rendered by the initial NeRF model into the training dataset. Through extensive experiments, we demonstrate that our method delivers highly competitive and robust performance in both mesh rendering quality and geometric quality.
Authors: Yuhong He, Long Peng, Qiaosi Yi, Chen Wu, Lu Wang
Abstract: Image restoration endeavors to reconstruct a high-quality, detail-rich image from a degraded counterpart, which is a pivotal process in photography and various computer vision systems. In real-world scenarios, different types of degradation can cause the loss of image details at various scales and degrade image contrast. Existing methods predominantly rely on CNN and Transformer to capture multi-scale representations. However, these methods are often limited by the high computational complexity of Transformers and the constrained receptive field of CNN, which hinder them from achieving superior performance and efficiency in image restoration. To address these challenges, we propose a novel Multi-Scale State-Space Model-based (MS-Mamba) for efficient image restoration that enhances the capacity for multi-scale representation learning through our proposed global and regional SSM modules. Additionally, an Adaptive Gradient Block (AGB) and a Residual Fourier Block (RFB) are proposed to improve the network's detail extraction capabilities by capturing gradients in various directions and facilitating learning details in the frequency domain. Extensive experiments on nine public benchmarks across four classic image restoration tasks, image deraining, dehazing, denoising, and low-light enhancement, demonstrate that our proposed method achieves new state-of-the-art performance while maintaining low computational complexity. The source code will be publicly available.
Authors: Shuxian Wang, Akshay Paruchuri, Zhaoxi Zhang, Sarah McGill, Roni Sengupta
Abstract: Monocular depth estimation in colonoscopy video aims to overcome the unusual lighting properties of the colonoscopic environment. One of the major challenges in this area is the domain gap between annotated but unrealistic synthetic data and unannotated but realistic clinical data. Previous attempts to bridge this domain gap directly target the depth estimation task itself. We propose a general pipeline of structure-preserving synthetic-to-real (sim2real) image translation (producing a modified version of the input image) to retain depth geometry through the translation process. This allows us to generate large quantities of realistic-looking synthetic images for supervised depth estimation with improved generalization to the clinical domain. We also propose a dataset of hand-picked sequences from clinical colonoscopies to improve the image translation process. We demonstrate the simultaneous realism of the translated images and preservation of depth maps via the performance of downstream depth estimation on various datasets.
Authors: Liyuan Zhu, Yue Li, Erik Sandstr\"om, Konrad Schindler, Iro Armeni
Abstract: Simultaneous Localization and Mapping (SLAM) based on 3D Gaussian Splats (3DGS) has recently shown promise towards more accurate, dense 3D scene maps. However, existing 3DGS-based methods fail to address the global consistency of the scene via loop closure and/or global bundle adjustment. To this end, we propose LoopSplat, which takes RGB-D images as input and performs dense mapping with 3DGS submaps and frame-to-model tracking. LoopSplat triggers loop closure online and computes relative loop edge constraints between submaps directly via 3DGS registration, leading to improvements in efficiency and accuracy over traditional global-to-local point cloud registration. It uses a robust pose graph optimization formulation and rigidly aligns the submaps to achieve global consistency. Evaluation on the synthetic Replica and real-world TUM-RGBD, ScanNet, and ScanNet++ datasets demonstrates competitive or superior tracking, mapping, and rendering compared to existing methods for dense RGB-D SLAM. Code is available at \href{https://loopsplat.github.io/}{loopsplat.github.io}.
Authors: Zhiyong Zhang, Aniket Gupta, Huaizu Jiang, Hanumant Singh
Abstract: Real-time high-accuracy optical flow estimation is crucial for various real-world applications. While recent learning-based optical flow methods have achieved high accuracy, they often come with significant computational costs. In this paper, we propose a highly efficient optical flow method that balances high accuracy with reduced computational demands. Building upon NeuFlow v1, we introduce new components including a much more light-weight backbone and a fast refinement module. Both these modules help in keeping the computational demands light while providing close to state of the art accuracy. Compares to other state of the art methods, our model achieves a 10x-70x speedup while maintaining comparable performance on both synthetic and real-world data. It is capable of running at over 20 FPS on 512x384 resolution images on a Jetson Orin Nano. The full training and evaluation code is available at https://github.com/neufieldrobotics/NeuFlow_v2.
Authors: Rafael M. Mamede, Pedro C. Neto, Ana F. Sequeira
Abstract: This study investigates the effects of occlusions on the fairness of face recognition systems, particularly focusing on demographic biases. Using the Racial Faces in the Wild (RFW) dataset and synthetically added realistic occlusions, we evaluate their effect on the performance of face recognition models trained on the BUPT-Balanced and BUPT-GlobalFace datasets. We note increases in the dispersion of FMR, FNMR, and accuracy alongside decreases in fairness according to Equilized Odds, Demographic Parity, STD of Accuracy, and Fairness Discrepancy Rate. Additionally, we utilize a pixel attribution method to understand the importance of occlusions in model predictions, proposing a new metric, Face Occlusion Impact Ratio (FOIR), that quantifies the extent to which occlusions affect model performance across different demographic groups. Our results indicate that occlusions exacerbate existing demographic biases, with models placing higher importance on occlusions in an unequal fashion, particularly affecting African individuals more severely.
Authors: Yifan Wang, Di Huang, Weicai Ye, Guofeng Zhang, Wanli Ouyang, Tong He
Abstract: Signed Distance Function (SDF)-based volume rendering has demonstrated significant capabilities in surface reconstruction. Although promising, SDF-based methods often fail to capture detailed geometric structures, resulting in visible defects. By comparing SDF-based volume rendering to density-based volume rendering, we identify two main factors within the SDF-based approach that degrade surface quality: SDF-to-density representation and geometric regularization. These factors introduce challenges that hinder the optimization of the SDF field. To address these issues, we introduce NeuRodin, a novel two-stage neural surface reconstruction framework that not only achieves high-fidelity surface reconstruction but also retains the flexible optimization characteristics of density-based methods. NeuRodin incorporates innovative strategies that facilitate transformation of arbitrary topologies and reduce artifacts associated with density bias. Extensive evaluations on the Tanks and Temples and ScanNet++ datasets demonstrate the superiority of NeuRodin, showing strong reconstruction capabilities for both indoor and outdoor environments using solely posed RGB captures. Project website: https://open3dvlab.github.io/NeuRodin/
Authors: Rasha Alshawi, Md Meftahul Ferdaus, Mahdi Abdelguerfi, Kendall Niles, Ken Pathak, Steve Sloan
Abstract: Imbalanced datasets are a significant challenge in real-world scenarios. They lead to models that underperform on underrepresented classes, which is a critical issue in infrastructure inspection. This paper introduces the Enhanced Feature Pyramid Network (E-FPN), a deep learning model for the semantic segmentation of culverts and sewer pipes within imbalanced datasets. The E-FPN incorporates architectural innovations like sparsely connected blocks and depth-wise separable convolutions to improve feature extraction and handle object variations. To address dataset imbalance, the model employs strategies like class decomposition and data augmentation. Experimental results on the culvert-sewer defects dataset and a benchmark aerial semantic segmentation drone dataset show that the E-FPN outperforms state-of-the-art methods, achieving an average Intersection over Union (IoU) improvement of 13.8% and 27.2%, respectively. Additionally, class decomposition and data augmentation together boost the model's performance by approximately 6.9% IoU. The proposed E-FPN presents a promising solution for enhancing object segmentation in challenging, multi-class real-world datasets, with potential applications extending beyond culvert-sewer defect detection.
Authors: Muhammad Al\`i, Francesca Razzano, Sergio Vitale, Giampaolo Ferraioli, Vito Pascazio, Gilda Schirinzi, Silvia Ullo
Abstract: Typically, the detection of marine debris relies on in-situ campaigns that are characterized by huge human effort and limited spatial coverage. Following the need of a rapid solution for the detection of floating plastic, methods based on remote sensing data have been proposed recently. Their main limitation is represented by the lack of a general reference for evaluating performance. Recently, the Marine Debris Archive (MARIDA) has been released as a standard dataset to develop and evaluate Machine Learning (ML) algorithms for detection of Marine Plastic Debris. The MARIDA dataset has been created for simplifying the comparison between detection solutions with the aim of stimulating the research in the field of marine environment preservation. In this work, an assessment of spectral based solutions is proposed by evaluating performance on MARIDA dataset. The outcome highlights the need of precise reference for fair evaluation.
Authors: Fuzhao Xue, Yukang Chen, Dacheng Li, Qinghao Hu, Ligeng Zhu, Xiuyu Li, Yunhao Fang, Haotian Tang, Shang Yang, Zhijian Liu, Ethan He, Hongxu Yin, Pavlo Molchanov, Jan Kautz, Linxi Fan, Yuke Zhu, Yao Lu, Song Han
Abstract: Long-context capability is critical for multi-modal foundation models. We introduce LongVILA, a full-stack solution for long-context vision-language models, including system, model training, and dataset development. On the system side, we introduce the first Multi-Modal Sequence Parallelism (MM-SP) system that enables long-context training and inference, enabling 2M context length training on 256 GPUs. MM-SP is also efficient, being 2.1x - 5.7x faster than Ring-Style Sequence Parallelism and 1.1x - 1.4x faster than Megatron-LM in text-only settings. Moreover, it seamlessly integrates with Hugging Face Transformers. For model training, we propose a five-stage pipeline comprising alignment, pre-training, context extension, and long-short joint supervised fine-tuning. Regarding datasets, we meticulously construct large-scale visual language pre-training datasets and long video instruction-following datasets to support our multi-stage training process. The full-stack solution extends the feasible frame number of VILA by a factor of 128 (from 8 to 1024 frames) and improves long video captioning score from 2.00 to 3.26 (1.6x), achieving 99.5% accuracy in 1400-frames video (274k context length) needle in a haystack. LongVILA-8B also demonstrates a consistent improvement in performance on long videos within the VideoMME benchmark as the video frames increase.
Authors: Chao Xu, Ang Li, Linghao Chen, Yulin Liu, Ruoxi Shi, Hao Su, Minghua Liu
Abstract: Open-world 3D generation has recently attracted considerable attention. While many single-image-to-3D methods have yielded visually appealing outcomes, they often lack sufficient controllability and tend to produce hallucinated regions that may not align with users' expectations. In this paper, we explore an important scenario in which the input consists of one or a few unposed 2D images of a single object, with little or no overlap. We propose a novel method, SpaRP, to reconstruct a 3D textured mesh and estimate the relative camera poses for these sparse-view images. SpaRP distills knowledge from 2D diffusion models and finetunes them to implicitly deduce the 3D spatial relationships between the sparse views. The diffusion model is trained to jointly predict surrogate representations for camera poses and multi-view images of the object under known poses, integrating all information from the input sparse views. These predictions are then leveraged to accomplish 3D reconstruction and pose estimation, and the reconstructed 3D model can be used to further refine the camera poses of input views. Through extensive experiments on three datasets, we demonstrate that our method not only significantly outperforms baseline methods in terms of 3D reconstruction quality and pose prediction accuracy but also exhibits strong efficiency. It requires only about 20 seconds to produce a textured mesh and camera poses for the input views. Project page: https://chaoxu.xyz/sparp.
Authors: Minghua Liu, Chong Zeng, Xinyue Wei, Ruoxi Shi, Linghao Chen, Chao Xu, Mengqi Zhang, Zhaoning Wang, Xiaoshuai Zhang, Isabella Liu, Hongzhi Wu, Hao Su
Abstract: Open-world 3D reconstruction models have recently garnered significant attention. However, without sufficient 3D inductive bias, existing methods typically entail expensive training costs and struggle to extract high-quality 3D meshes. In this work, we introduce MeshFormer, a sparse-view reconstruction model that explicitly leverages 3D native structure, input guidance, and training supervision. Specifically, instead of using a triplane representation, we store features in 3D sparse voxels and combine transformers with 3D convolutions to leverage an explicit 3D structure and projective bias. In addition to sparse-view RGB input, we require the network to take input and generate corresponding normal maps. The input normal maps can be predicted by 2D diffusion models, significantly aiding in the guidance and refinement of the geometry's learning. Moreover, by combining Signed Distance Function (SDF) supervision with surface rendering, we directly learn to generate high-quality meshes without the need for complex multi-stage training processes. By incorporating these explicit 3D biases, MeshFormer can be trained efficiently and deliver high-quality textured meshes with fine-grained geometric details. It can also be integrated with 2D diffusion models to enable fast single-image-to-3D and text-to-3D tasks. Project page: https://meshformer3d.github.io
Authors: Yusuke Hirota, Min-Hung Chen, Chien-Yi Wang, Yuta Nakashima, Yu-Chiang Frank Wang, Ryo Hachiuma
Abstract: Large-scale vision-language models, such as CLIP, are known to contain harmful societal bias regarding protected attributes (e.g., gender and age). In this paper, we aim to address the problems of societal bias in CLIP. Although previous studies have proposed to debias societal bias through adversarial learning or test-time projecting, our comprehensive study of these works identifies two critical limitations: 1) loss of attribute information when it is explicitly disclosed in the input and 2) use of the attribute annotations during debiasing process. To mitigate societal bias in CLIP and overcome these limitations simultaneously, we introduce a simple-yet-effective debiasing method called SANER (societal attribute neutralizer) that eliminates attribute information from CLIP text features only of attribute-neutral descriptions. Experimental results show that SANER, which does not require attribute annotations and preserves original information for attribute-specific descriptions, demonstrates superior debiasing ability than the existing methods.
Authors: Li Qiao, Mahdi Boloursaz Mashhadi, Zhen Gao, Chuan Heng Foh, Pei Xiao, Mehdi Bennis
Abstract: Generative foundation AI models have recently shown great success in synthesizing natural signals with high perceptual quality using only textual prompts and conditioning signals to guide the generation process. This enables semantic communications at extremely low data rates in future wireless networks. In this paper, we develop a latency-aware semantic communications framework with pre-trained generative models. The transmitter performs multi-modal semantic decomposition on the input signal and transmits each semantic stream with the appropriate coding and communication schemes based on the intent. For the prompt, we adopt a re-transmission-based scheme to ensure reliable transmission, and for the other semantic modalities we use an adaptive modulation/coding scheme to achieve robustness to the changing wireless channel. Furthermore, we design a semantic and latency-aware scheme to allocate transmission power to different semantic modalities based on their importance subjected to semantic quality constraints. At the receiver, a pre-trained generative model synthesizes a high fidelity signal using the received multi-stream semantics. Simulation results demonstrate ultra-low-rate, low-latency, and channel-adaptive semantic communications.
Authors: Zhili Cheng, Zhitong Wang, Jinyi Hu, Shengding Hu, An Liu, Yuge Tu, Pengkai Li, Lei Shi, Zhiyuan Liu, Maosong Sun
Abstract: Despite advancements in Large Language Models (LLMs) and Large Multimodal Models (LMMs), their integration into language-grounded, human-like embodied agents remains incomplete, hindering complex real-life task performance in physical environments. Existing integrations often feature limited open sourcing, challenging collective progress in this field. We introduce LEGENT, an open, scalable platform for developing embodied agents using LLMs and LMMs. LEGENT offers a dual approach: a rich, interactive 3D environment with communicable and actionable agents, paired with a user-friendly interface, and a sophisticated data generation pipeline utilizing advanced algorithms to exploit supervision from simulated worlds at scale. In our experiments, an embryonic vision-language-action model trained on LEGENT-generated data surpasses GPT-4V in embodied tasks, showcasing promising generalization capabilities.
Authors: Xin Wang, Xiaoyu Liu, Peng Huang, Pu Huang, Shu Hu, Hongtu Zhu
Abstract: Medical Image Foundation Models have proven to be powerful tools for mask prediction across various datasets. However, accurately assessing the uncertainty of their predictions remains a significant challenge. To address this, we propose a new model, U-MedSAM, which integrates the MedSAM model with an uncertainty-aware loss function and the Sharpness-Aware Minimization (SharpMin) optimizer. The uncertainty-aware loss function automatically combines region-based, distribution-based, and pixel-based loss designs to enhance segmentation accuracy and robustness. SharpMin improves generalization by finding flat minima in the loss landscape, thereby reducing overfitting. Our method was evaluated in the CVPR24 MedSAM on Laptop challenge, where U-MedSAM demonstrated promising performance.
Authors: Florian Mouret (CESBIO, UO), David Morin (CESBIO), Milena Planells (CESBIO), C\'ecile Vincent-Barbaroux
Abstract: This paper investigates tree species classification using Sentinel-2 multispectral satellite image time-series. Despite their critical importance for many applications, such maps are often unavailable, outdated, or inaccurate for large areas. The interest of using remote sensing time series to produce these maps has been highlighted in many studies. However, many methods proposed in the literature still rely on a standard classification algorithm, usually the Random Forest (RF) algorithm with vegetation indices. This study shows that the use of deep learning models can lead to a significant improvement in classification results, especially in an imbalanced context where the RF algorithm tends to predict towards the majority class. In our use case in the center of France with 10 tree species, we obtain an overall accuracy (OA) around 95% and a F1-macro score around 80% using three different benchmark deep learning architectures. In contrast, using the RF algorithm yields an OA of 93% and an F1 of 60%, indicating that the minority classes are not classified with sufficient accuracy. Therefore, the proposed framework is a strong baseline that can be easily implemented in most scenarios, even with a limited amount of reference data. Our results highlight that standard multilayer perceptron can be competitive with batch normalization and a sufficient amount of parameters. Other architectures (convolutional or attention-based) can also achieve strong results when tuned properly. Furthermore, our results show that DL models are naturally robust to imbalanced data, although similar results can be obtained using dedicated techniques.
Authors: Lingxin Jin, Xianyu Wen, Wei Jiang, Jinyu Zhan
Abstract: Deep Neural Networks (DNNs) have found extensive applications in safety-critical artificial intelligence systems, such as autonomous driving and facial recognition systems. However, recent research has revealed their susceptibility to Neural Network Trojans (NN Trojans) maliciously injected by adversaries. This vulnerability arises due to the intricate architecture and opacity of DNNs, resulting in numerous redundant neurons embedded within the models. Adversaries exploit these vulnerabilities to conceal malicious Trojans within DNNs, thereby causing erroneous outputs and posing substantial threats to the efficacy of DNN-based applications. This article presents a comprehensive survey of Trojan attacks against DNNs and the countermeasure methods employed to mitigate them. Initially, we trace the evolution of the concept from traditional Trojans to NN Trojans, highlighting the feasibility and practicality of generating NN Trojans. Subsequently, we provide an overview of notable works encompassing various attack and defense strategies, facilitating a comparative analysis of their approaches. Through these discussions, we offer constructive insights aimed at refining these techniques. In recognition of the gravity and immediacy of this subject matter, we also assess the feasibility of deploying such attacks in real-world scenarios as opposed to controlled ideal datasets. The potential real-world implications underscore the urgency of addressing this issue effectively.
Authors: Qingyuan Zeng, Zhenzhong Wang, Yiu-ming Cheung, Min Jiang
Abstract: While image-to-text models have demonstrated significant advancements in various vision-language tasks, they remain susceptible to adversarial attacks. Existing white-box attacks on image-to-text models require access to the architecture, gradients, and parameters of the target model, resulting in low practicality. Although the recently proposed gray-box attacks have improved practicality, they suffer from semantic loss during the training process, which limits their targeted attack performance. To advance adversarial attacks of image-to-text models, this paper focuses on a challenging scenario: decision-based black-box targeted attacks where the attackers only have access to the final output text and aim to perform targeted attacks. Specifically, we formulate the decision-based black-box targeted attack as a large-scale optimization problem. To efficiently solve the optimization problem, a three-stage process \textit{Ask, Attend, Attack}, called \textit{AAA}, is proposed to coordinate with the solver. \textit{Ask} guides attackers to create target texts that satisfy the specific semantics. \textit{Attend} identifies the crucial regions of the image for attacking, thus reducing the search space for the subsequent \textit{Attack}. \textit{Attack} uses an evolutionary algorithm to attack the crucial regions, where the attacks are semantically related to the target texts of \textit{Ask}, thus achieving targeted attacks without semantic loss. Experimental results on transformer-based and CNN+RNN-based image-to-text models confirmed the effectiveness of our proposed \textit{AAA}.
Authors: Arwen Bradley, Preetum Nakkiran
Abstract: We investigate the theoretical foundations of classifier-free guidance (CFG). CFG is the dominant method of conditional sampling for text-to-image diffusion models, yet unlike other aspects of diffusion, it remains on shaky theoretical footing. In this paper, we disprove common misconceptions, by showing that CFG interacts differently with DDPM (Ho et al., 2020) and DDIM (Song et al., 2021), and neither sampler with CFG generates the gamma-powered distribution $p(x|c)^\gamma p(x)^{1-\gamma}$. Then, we clarify the behavior of CFG by showing that it is a kind of predictor-corrector method (Song et al., 2020) that alternates between denoising and sharpening, which we call predictor-corrector guidance (PCG). We prove that in the SDE limit, CFG is actually equivalent to combining a DDIM predictor for the conditional distribution together with a Langevin dynamics corrector for a gamma-powered distribution (with a carefully chosen gamma). Our work thus provides a lens to theoretically understand CFG by embedding it in a broader design space of principled sampling methods.
Authors: Lin Zuo, Yongqi Ding, Wenwei Luo, Mengmeng Jing, Xianlong Tian, Kunshan Yang
Abstract: Spiking neural networks (SNNs) have received widespread attention as an ultra-low energy computing paradigm. Recent studies have focused on improving the feature extraction capability of SNNs, but they suffer from inefficient inference and suboptimal performance. In this paper, we propose a simple yet effective temporal reversed training (TRT) method to optimize the spatio-temporal performance of SNNs and circumvent these problems. We perturb the input temporal data by temporal reversal, prompting the SNN to produce original-reversed consistent output logits and to learn perturbation-invariant representations. For static data without temporal dimension, we generalize this strategy by exploiting the inherent temporal property of spiking neurons for spike feature temporal reversal. In addition, we utilize the lightweight ``star operation" (element-wise multiplication) to hybridize the original and temporally reversed spike firing rates and expand the implicit dimensions, which serves as spatio-temporal regularization to further enhance the generalization of the SNN. Our method involves only an additional temporal reversal operation and element-wise multiplication during training, thus incurring negligible training overhead and not affecting the inference efficiency at all. Extensive experiments on static/neuromorphic object/action recognition, and 3D point cloud classification tasks demonstrate the effectiveness and generalizability of our method. In particular, with only two timesteps, our method achieves 74.77\% and 90.57\% accuracy on ImageNet and ModelNet40, respectively.
Authors: Jaehyuk Lim, Bruce W. Lee
Abstract: This paper introduces and examines the phenomenon of "visual sycophancy" in multimodal language models, a term we propose to describe these models' tendency to disproportionately favor visually presented information, even when it contradicts their prior knowledge or responses. Our study employs a systematic methodology to investigate this phenomenon: we present models with images of multiple-choice questions, which they initially answer correctly, then expose the same model to versions with visually pre-marked options. Our findings reveal a significant shift in the models' responses towards the pre-marked option despite their previous correct answers. Comprehensive evaluations demonstrate that visual sycophancy is a consistent and quantifiable behavior across various model architectures. Our findings highlight potential limitations in the reliability of these models when processing potentially misleading visual information, raising important questions about their application in critical decision-making contexts.
Authors: SeungHyun Kim, Seohyeon Jung, Seonghyeon Kim, Juho Lee
Abstract: Bayesian Neural Networks(BNNs) with high-dimensional parameters pose a challenge for posterior inference due to the multi-modality of the posterior distributions. Stochastic Gradient MCMC(SGMCMC) with cyclical learning rate scheduling is a promising solution, but it requires a large number of sampling steps to explore high-dimensional multi-modal posteriors, making it computationally expensive. In this paper, we propose a meta-learning strategy to build \gls{sgmcmc} which can efficiently explore the multi-modal target distributions. Our algorithm allows the learned SGMCMC to quickly explore the high-density region of the posterior landscape. Also, we show that this exploration property is transferrable to various tasks, even for the ones unseen during a meta-training stage. Using popular image classification benchmarks and a variety of downstream tasks, we demonstrate that our method significantly improves the sampling efficiency, achieving better performance than vanilla \gls{sgmcmc} without incurring significant computational overhead.
Authors: Cho Yang
Abstract: CycleGAN was trained on SynthRAD Grand Challenge Dataset using the single-epoch modification (SEM) method proposed in this paper which is referred to as (CycleGAN-single) compared to the usual method of training CycleGAN on around 200 epochs (CycleGAN-multi). Model performance were evaluated qualitatively and quantitatively with quantitative performance metrics like PSNR, SSIM, MAE and MSE. The consideration of both quantitative and qualitative performance when evaluating a model is unique to certain image-translation tasks like medical imaging as detailed in this paper. Also, this paper shows that good quantitative performance does not always imply good qualitative performance and the converse is also not always True (i.e. good qualitative performance does not always imply good quantitative performance). This paper also proposes FQGA (Fast Paired Image-to-Image Translation Quarter-Generator Adversary) Model which has 1/4 the number of parameters compared to CycleGAN (when comparing their Generator Models). FQGA outperforms CycleGAN qualitatively and quantitatively even only after training on 20 epochs. Finally, using SEM method on FQGA allowed it to again outperform CycleGAN both quantitatively and qualitatively. These performance gains with fewer model parameters and time savings from running fewer epochs may also be applicable to other image-to-image translation tasks in Machine Learning apart from the Medical image-translation task discussed in this paper between Cone Beam Computed Tomography (CBCT) and Computed Tomography (CT) images.
Authors: Junchao Zhu, Mengmeng Yin, Ruining Deng, Yitian Long, Yu Wang, Yaohong Wang, Shilin Zhao, Haichun Yang, Yuankai Huo
Abstract: Accurate delineation of the boundaries between the renal cortex and medulla is crucial for subsequent functional structural analysis and disease diagnosis. Training high-quality deep-learning models for layer segmentation relies on the availability of large amounts of annotated data. However, due to the patient's privacy of medical data and scarce clinical cases, constructing pathological datasets from clinical sources is relatively difficult and expensive. Moreover, using external natural image datasets introduces noise during the domain generalization process. Cross-species homologous data, such as mouse kidney data, which exhibits high structural and feature similarity to human kidneys, has the potential to enhance model performance on human datasets. In this study, we incorporated the collected private Periodic Acid-Schiff (PAS) stained mouse kidney dataset into the human kidney dataset for joint training. The results showed that after introducing cross-species homologous data, the semantic segmentation models based on CNN and Transformer architectures achieved an average increase of 1.77% and 1.24% in mIoU, and 1.76% and 0.89% in Dice score for the human renal cortex and medulla datasets, respectively. This approach is also capable of enhancing the model's generalization ability. This indicates that cross-species homologous data, as a low-noise trainable data source, can help improve model performance under conditions of limited clinical samples. Code is available at https://github.com/hrlblab/layer_segmentation.
Authors: Mengqi Wu, Minhui Yu, Shuaiming Jing, Pew-Thian Yap, Zhengwu Zhang, Mingxia Liu
Abstract: Multi-site structural MRI is increasingly used in neuroimaging studies to diversify subject cohorts. However, combining MR images acquired from various sites/centers may introduce site-related non-biological variations. Retrospective image harmonization helps address this issue, but current methods usually perform harmonization on pre-extracted hand-crafted radiomic features, limiting downstream applicability. Several image-level approaches focus on 2D slices, disregarding inherent volumetric information, leading to suboptimal outcomes. To this end, we propose a novel 3D MRI Harmonization framework through Conditional Latent Diffusion (HCLD) by explicitly considering image style and brain anatomy. It comprises a generalizable 3D autoencoder that encodes and decodes MRIs through a 4D latent space, and a conditional latent diffusion model that learns the latent distribution and generates harmonized MRIs with anatomical information from source MRIs while conditioned on target image style. This enables efficient volume-level MRI harmonization through latent style translation, without requiring paired images from target and source domains during training. The HCLD is trained and evaluated on 4,158 T1-weighted brain MRIs from three datasets in three tasks, assessing its ability to remove site-related variations while retaining essential biological features. Qualitative and quantitative experiments suggest the effectiveness of HCLD over several state-of-the-arts
Authors: Xiawei Wang, James Sharpnack, Thomas C. M. Lee
Abstract: Lung cancer is a major cause of cancer-related deaths, and early diagnosis and treatment are crucial for improving patients' survival outcomes. In this paper, we propose to employ convolutional neural networks to model the non-linear relationship between the risk of lung cancer and the lungs' morphology revealed in the CT images. We apply a mini-batched loss that extends the Cox proportional hazards model to handle the non-convexity induced by neural networks, which also enables the training of large data sets. Additionally, we propose to combine mini-batched loss and binary cross-entropy to predict both lung cancer occurrence and the risk of mortality. Simulation results demonstrate the effectiveness of both the mini-batched loss with and without the censoring mechanism, as well as its combination with binary cross-entropy. We evaluate our approach on the National Lung Screening Trial data set with several 3D convolutional neural network architectures, achieving high AUC and C-index scores for lung cancer classification and survival prediction. These results, obtained from simulations and real data experiments, highlight the potential of our approach to improving the diagnosis and treatment of lung cancer.
Authors: Guoqing Zhang, Jingyun Yang, Yang Li
Abstract: As the rapid development of computer vision and the emergence of powerful network backbones and architectures, the application of deep learning in medical imaging has become increasingly significant. Unlike natural images, medical images lack huge volumes of data but feature more modalities, making it difficult to train a general model that has satisfactory performance across various datasets. In practice, practitioners often suffer from manually creating and testing models combining independent backbones and architectures, which is a laborious and time-consuming process. We propose Flemme, a FLExible and Modular learning platform for MEdical images. Our platform separates encoders from the model architectures so that different models can be constructed via various combinations of supported encoders and architectures. We construct encoders using building blocks based on convolution, transformer, and state-space model (SSM) to process both 2D and 3D image patches. A base architecture is implemented following an encoder-decoder style, with several derived architectures for image segmentation, reconstruction, and generation tasks. In addition, we propose a general hierarchical architecture incorporating a pyramid loss to optimize and fuse vertical features. Experiments demonstrate that this simple design leads to an average improvement of 5.60% in Dice score and 7.81% in mean interaction of units (mIoU) for segmentation models, as well as an enhancement of 5.57% in peak signal-to-noise ratio (PSNR) and 8.22% in structural similarity (SSIM) for reconstruction models. We further utilize Flemme as an analytical tool to assess the effectiveness and efficiency of various encoders across different tasks. Code is available at https://github.com/wlsdzyzl/flemme.
Authors: Jiahao Su, Kang You, Zekai Xu, Weizhi Xu, Zhezhi He
Abstract: Spiking neural networks (SNNs) are becoming a promising alternative to conventional artificial neural networks (ANNs) due to their rich neural dynamics and the implementation of energy-efficient neuromorphic chips. However, the non-differential binary communication mechanism makes SNN hard to converge to an ANN-level accuracy. When SNN encounters sequence learning, the situation becomes worse due to the difficulties in modeling long-range dependencies. To overcome these difficulties, researchers developed variants of LIF neurons and different surrogate gradients but still failed to obtain good results when the sequence became longer (e.g., $>$500). Unlike them, we obtain an optimal SNN in sequence learning by directly mapping parameters from a quantized CRNN. We design two sub-pipelines to support the end-to-end conversion of different structures in neural networks, which is called CNN-Morph (CNN $\rightarrow$ QCNN $\rightarrow$ BIFSNN) and RNN-Morph (RNN $\rightarrow$ QRNN $\rightarrow$ RBIFSNN). Using conversion pipelines and the s-analog encoding method, the conversion error of our framework is zero. Furthermore, we give the theoretical and experimental demonstration of the lossless CRNN-SNN conversion. Our results show the effectiveness of our method over short and long timescales tasks compared with the state-of-the-art learning- and conversion-based methods. We reach the highest accuracy of 99.16% (0.46 $\uparrow$) on S-MNIST, 94.95% (3.95 $\uparrow$) on PS-MNIST (sequence length of 784) respectively, and the lowest loss of 0.057 (0.013 $\downarrow$) within 8 time-steps in collision avoidance dataset.
Authors: Bowen Xin, Tony Young, Claire E Wainwright, Tamara Blake, Leo Lebrat, Thomas Gaass, Thomas Benkert, Alto Stemmer, David Coman, Jason Dowling
Abstract: Medical image synthesis generates additional imaging modalities that are costly, invasive or harmful to acquire, which helps to facilitate the clinical workflow. When training pairs are substantially misaligned (e.g., lung MRI-CT pairs with respiratory motion), accurate image synthesis remains a critical challenge. Recent works explored the directional registration module to adjust misalignment in generative adversarial networks (GANs); however, substantial misalignment will lead to 1) suboptimal data mapping caused by correspondence ambiguity, and 2) degraded image fidelity caused by morphology influence on discriminators. To address the challenges, we propose a novel Deformation-aware GAN (DA-GAN) to dynamically correct the misalignment during the image synthesis based on multi-objective inverse consistency. Specifically, in the generative process, three levels of inverse consistency cohesively optimise symmetric registration and image generation for improved correspondence. In the adversarial process, to further improve image fidelity under misalignment, we design deformation-aware discriminators to disentangle the mismatched spatial morphology from the judgement of image fidelity. Experimental results show that DA-GAN achieved superior performance on a public dataset with simulated misalignments and a real-world lung MRI-CT dataset with respiratory motion misalignment. The results indicate the potential for a wide range of medical image synthesis tasks such as radiotherapy planning.
Authors: Yi Liu, Junchen Ding, Gelei Deng, Yuekang Li, Tianwei Zhang, Weisong Sun, Yaowen Zheng, Jingquan Ge, Yang Liu
Abstract: Geolocation is now a vital aspect of modern life, offering numerous benefits but also presenting serious privacy concerns. The advent of large vision-language models (LVLMs) with advanced image-processing capabilities introduces new risks, as these models can inadvertently reveal sensitive geolocation information. This paper presents the first in-depth study analyzing the challenges posed by traditional deep learning and LVLM-based geolocation methods. Our findings reveal that LVLMs can accurately determine geolocations from images, even without explicit geographic training. To address these challenges, we introduce \tool{}, an innovative framework that significantly enhances image-based geolocation accuracy. \tool{} employs a systematic chain-of-thought (CoT) approach, mimicking human geoguessing strategies by carefully analyzing visual and contextual cues such as vehicle types, architectural styles, natural landscapes, and cultural elements. Extensive testing on a dataset of 50,000 ground-truth data points shows that \tool{} outperforms both traditional models and human benchmarks in accuracy. It achieves an impressive average score of 4550.5 in the GeoGuessr game, with an 85.37\% win rate, and delivers highly precise geolocation predictions, with the closest distances as accurate as 0.3 km. Furthermore, our study highlights issues related to dataset integrity, leading to the creation of a more robust dataset and a refined framework that leverages LVLMs' cognitive capabilities to improve geolocation precision. These findings underscore \tool{}'s superior ability to interpret complex visual data, the urgent need to address emerging security vulnerabilities posed by LVLMs, and the importance of responsible AI development to ensure user privacy protection.
Authors: Yi Kan Wang, Ludmila Tydlitatova, Jeremy D. Kunz, Gerard Oakley, Ran A. Godrich, Matthew C. H. Lee, Chad Vanderbilt, Razik Yousfi, Thomas Fuchs, David S. Klimstra, Siqi Liu
Abstract: Many molecular alterations serve as clinically prognostic or therapy-predictive biomarkers, typically detected using single or multi-gene molecular assays. However, these assays are expensive, tissue destructive and often take weeks to complete. Using AI on routine H&E WSIs offers a fast and economical approach to screen for multiple molecular biomarkers. We present a high-throughput AI-based system leveraging Virchow2, a foundation model pre-trained on 3 million slides, to interrogate genomic features previously determined by an next-generation sequencing (NGS) assay, using 47,960 scanned hematoxylin and eosin (H&E) whole slide images (WSIs) from 38,984 cancer patients. Unlike traditional methods that train individual models for each biomarker or cancer type, our system employs a unified model to simultaneously predict a wide range of clinically relevant molecular biomarkers across cancer types. By training the network to replicate the MSK-IMPACT targeted biomarker panel of 505 genes, it identified 80 high performing biomarkers with a mean AU-ROC of 0.89 in 15 most common cancer types. In addition, 40 biomarkers demonstrated strong associations with specific cancer histologic subtypes. Furthermore, 58 biomarkers were associated with targets frequently assayed clinically for therapy selection and response prediction. The model can also predict the activity of five canonical signaling pathways, identify defects in DNA repair mechanisms, and predict genomic instability measured by tumor mutation burden, microsatellite instability (MSI), and chromosomal instability (CIN). The proposed model can offer potential to guide therapy selection, improve treatment efficacy, accelerate patient screening for clinical trials and provoke the interrogation of new therapeutic targets.
Authors: Shahzaib Iqbal, Muhammad Zeeshan, Mehwish Mehmood, Tariq M. Khan, Imran Razzak
Abstract: Early detection of skin cancer relies on precise segmentation of dermoscopic images of skin lesions. However, this task is challenging due to the irregular shape of the lesion, the lack of sharp borders, and the presence of artefacts such as marker colours and hair follicles. Recent methods for melanoma segmentation are U-Nets and fully connected networks (FCNs). As the depth of these neural network models increases, they can face issues like the vanishing gradient problem and parameter redundancy, potentially leading to a decrease in the Jaccard index of the segmentation model. In this study, we introduced a novel network named TESL-Net for the segmentation of skin lesions. The proposed TESL-Net involves a hybrid network that combines the local features of a CNN encoder-decoder architecture with long-range and temporal dependencies using bi-convolutional long-short-term memory (Bi-ConvLSTM) networks and a Swin transformer. This enables the model to account for the uncertainty of segmentation over time and capture contextual channel relationships in the data. We evaluated the efficacy of TESL-Net in three commonly used datasets (ISIC 2016, ISIC 2017, and ISIC 2018) for the segmentation of skin lesions. The proposed TESL-Net achieves state-of-the-art performance, as evidenced by a significantly elevated Jaccard index demonstrated by empirical results.
Authors: Zhi Qiao, Linbin Han, Xiantong Zhen, Jia-Hong Gao, Zhen Qian
Abstract: In light of the inherent entailment relations between images and text, hyperbolic point vector embeddings, leveraging the hierarchical modeling advantages of hyperbolic space, have been utilized for visual semantic representation learning. However, point vector embedding approaches fail to address the issue of semantic uncertainty, where an image may have multiple interpretations, and text may refer to different images, a phenomenon particularly prevalent in the medical domain. Therefor, we propose \textbf{HYDEN}, a novel hyperbolic density embedding based image-text representation learning approach tailored for specific medical domain data. This method integrates text-aware local features alongside global features from images, mapping image-text features to density features in hyperbolic space via using hyperbolic pseudo-Gaussian distributions. An encapsulation loss function is employed to model the partial order relations between image-text density distributions. Experimental results demonstrate the interpretability of our approach and its superior performance compared to the baseline methods across various zero-shot tasks and different datasets.
Authors: Zhi Qiao, Xuhui Liu, Xiaopeng Wang, Runkun Liu, Xiantong Zhen, Pei Dong, Zhen Qian
Abstract: Intraoperative CT imaging serves as a crucial resource for surgical guidance; however, it may not always be readily accessible or practical to implement. In scenarios where CT imaging is not an option, reconstructing CT scans from X-rays can offer a viable alternative. In this paper, we introduce an innovative method for 3D CT reconstruction utilizing biplanar X-rays. Distinct from previous research that relies on conventional image generation techniques, our approach leverages a conditional diffusion process to tackle the task of reconstruction. More precisely, we employ a diffusion-based probabilistic model trained to produce 3D CT images based on orthogonal biplanar X-rays. To improve the structural integrity of the reconstructed images, we incorporate a novel projection loss function. Experimental results validate that our proposed method surpasses existing state-of-the-art benchmarks in both visual image quality and multiple evaluative metrics. Specifically, our technique achieves a higher Structural Similarity Index (SSIM) of 0.83, a relative increase of 10\%, and a lower Fr\'echet Inception Distance (FID) of 83.43, which represents a relative decrease of 25\%.
Authors: Zhi Qiao, Hanqiang Ouyang, Dongheng Chu, Huishu Yuan, Xiantong Zhen, Pei Dong, Zhen Qian
Abstract: For surgical planning and intra-operation imaging, CT reconstruction using X-ray images can potentially be an important alternative when CT imaging is not available or not feasible. In this paper, we aim to use biplanar X-rays to reconstruct a 3D CT image, because biplanar X-rays convey richer information than single-view X-rays and are more commonly used by surgeons. Different from previous studies in which the two X-ray views were treated indifferently when fusing the cross-view data, we propose a novel attention-informed coarse-to-fine cross-view fusion method to combine the features extracted from the orthogonal biplanar views. This method consists of a view attention alignment sub-module and a fine-distillation sub-module that are designed to work together to highlight the unique or complementary information from each of the views. Experiments have demonstrated the superiority of our proposed method over the SOTA methods.
Authors: Yunxin Li, Haoyuan Shi, Baotian Hu, Longyue Wang, Jiashun Zhu, Jinyi Xu, Zhen Zhao, Min Zhang
Abstract: Traditional animation generation methods depend on training generative models with human-labelled data, entailing a sophisticated multi-stage pipeline that demands substantial human effort and incurs high training costs. Due to limited prompting plans, these methods typically produce brief, information-poor, and context-incoherent animations. To overcome these limitations and automate the animation process, we pioneer the introduction of large multimodal models (LMMs) as the core processor to build an autonomous animation-making agent, named Anim-Director. This agent mainly harnesses the advanced understanding and reasoning capabilities of LMMs and generative AI tools to create animated videos from concise narratives or simple instructions. Specifically, it operates in three main stages: Firstly, the Anim-Director generates a coherent storyline from user inputs, followed by a detailed director's script that encompasses settings of character profiles and interior/exterior descriptions, and context-coherent scene descriptions that include appearing characters, interiors or exteriors, and scene events. Secondly, we employ LMMs with the image generation tool to produce visual images of settings and scenes. These images are designed to maintain visual consistency across different scenes using a visual-language prompting method that combines scene descriptions and images of the appearing character and setting. Thirdly, scene images serve as the foundation for producing animated videos, with LMMs generating prompts to guide this process. The whole process is notably autonomous without manual intervention, as the LMMs interact seamlessly with generative tools to generate prompts, evaluate visual quality, and select the best one to optimize the final output.
Authors: Jaejun Lee, Yoori Oh, Injune Hwang, Kyogu Lee
Abstract: This paper delves into the emerging field of face-based voice conversion, leveraging the unique relationship between an individual's facial features and their vocal characteristics. We present a novel face-based voice conversion framework that particularly utilizes the average fundamental frequency of the target speaker, derived solely from their facial images. Through extensive analysis, our framework demonstrates superior speech generation quality and the ability to align facial features with voice characteristics, including tracking of the target speaker's fundamental frequency.
Authors: Christoph Auer, Maksym Lysak, Ahmed Nassar, Michele Dolfi, Nikolaos Livathinos, Panos Vagenas, Cesar Berrospi Ramis, Matteo Omenetti, Fabian Lindlbauer, Kasper Dinkla, Valery Weber, Lucas Morin, Ingmar Meijer, Viktor Kuropiatnyk, Peter W. J. Staar
Abstract: This technical report introduces Docling, an easy to use, self-contained, MIT-licensed open-source package for PDF document conversion. It is powered by state-of-the-art specialized AI models for layout analysis (DocLayNet) and table structure recognition (TableFormer), and runs efficiently on commodity hardware in a small resource budget. The code interface allows for easy extensibility and addition of new features and models.
Authors: Silvia Seidlitz, Katharina H\"olzl, Ayca von Garrel, Jan Sellner, Stephan Katzenschlager, Tobias H\"olle, Dania Fischer, Maik von der Forst, Felix C. F. Schmitt, Markus A. Weigand, Lena Maier-Hein, Maximilian Dietrich
Abstract: With sepsis remaining a leading cause of mortality, early identification of septic patients and those at high risk of death is a challenge of high socioeconomic importance. The driving hypothesis of this study was that hyperspectral imaging (HSI) could provide novel biomarkers for sepsis diagnosis and treatment management due to its potential to monitor microcirculatory alterations. We conducted a comprehensive study involving HSI data of the palm and fingers from more than 480 patients on the day of their intensive care unit (ICU) admission. The findings demonstrate that HSI measurements can predict sepsis with an area under the receiver operating characteristic curve (AUROC) of 0.80 (95 % confidence interval (CI) [0.76; 0.84]) and mortality with an AUROC of 0.72 (95 % CI [0.65; 0.79]). The predictive performance improves substantially when additional clinical data is incorporated, leading to an AUROC of up to 0.94 (95 % CI [0.92; 0.96]) for sepsis and 0.84 (95 % CI [0.78; 0.89]) for mortality. We conclude that HSI presents novel imaging biomarkers for the rapid, non-invasive prediction of sepsis and mortality, suggesting its potential as an important modality for guiding diagnosis and treatment.
Authors: Chris Hyunchul Jo, Jiwoong Yang, Byunghwan Jeon, Hackjoon Shim, Ikbeom Jang
Abstract: Research question: We test whether a plane shoulder radiograph can be used together with deep learning methods to identify patients with rotator cuff tears as opposed to using an MRI in standard of care. Findings: By integrating convolutional block attention modules into a deep neural network, our model demonstrates high accuracy in detecting patients with rotator cuff tears, achieving an average AUC of 0.889 and an accuracy of 0.831. Meaning: This study validates the efficacy of our deep learning model to accurately detect rotation cuff tears from radiographs, offering a viable pre-assessment or alternative to more expensive imaging techniques such as MRI.
Authors: Weiji Kong, Xun Gong, Juan Wang
Abstract: Explaining the decisions of Deep Neural Networks (DNNs) for medical images has become increasingly important. Existing attribution methods have difficulty explaining the meaning of pixels while existing concept-based methods are limited by additional annotations or specific model structures that are difficult to apply to ultrasound images. In this paper, we propose the Lesion Concept Explainer (LCE) framework, which combines attribution methods with concept-based methods. We introduce the Segment Anything Model (SAM), fine-tuned on a large number of medical images, for concept discovery to enable a meaningful explanation of ultrasound image DNNs. The proposed framework is evaluated in terms of both faithfulness and understandability. We point out deficiencies in the popular faithfulness evaluation metrics and propose a new evaluation metric. Our evaluation of public and private breast ultrasound datasets (BUSI and FG-US-B) shows that LCE performs well compared to commonly-used explainability methods. Finally, we also validate that LCE can consistently provide reliable explanations for more meaningful fine-grained diagnostic tasks in breast ultrasound.
Authors: Hongyuan Zhang, Yanchen Xu, Sida Huang, Xuelong Li
Abstract: Inspired by the idea of Positive-incentive Noise (Pi-Noise or $\pi$-Noise) that aims at learning the reliable noise beneficial to tasks, we scientifically investigate the connection between contrastive learning and $\pi$-noise in this paper. By converting the contrastive loss to an auxiliary Gaussian distribution to quantitatively measure the difficulty of the specific contrastive model under the information theory framework, we properly define the task entropy, the core concept of $\pi$-noise, of contrastive learning. It is further proved that the predefined data augmentation in the standard contrastive learning paradigm can be regarded as a kind of point estimation of $\pi$-noise. Inspired by the theoretical study, a framework that develops a $\pi$-noise generator to learn the beneficial noise (instead of estimation) as data augmentations for contrast is proposed. The designed framework can be applied to diverse types of data and is also completely compatible with the existing contrastive models. From the visualization, we surprisingly find that the proposed method successfully learns effective augmentations.
Authors: Qianhui Men, Xiaoqing Guo, Aris T. Papageorghiou, J. Alison Noble
Abstract: 3D pose estimation from a 2D cross-sectional view enables healthcare professionals to navigate through the 3D space, and such techniques initiate automatic guidance in many image-guided radiology applications. In this work, we investigate how estimating 3D fetal pose from freehand 2D ultrasound scanning can guide a sonographer to locate a head standard plane. Fetal head pose is estimated by the proposed Pose-GuideNet, a novel 2D/3D registration approach to align freehand 2D ultrasound to a 3D anatomical atlas without the acquisition of 3D ultrasound. To facilitate the 2D to 3D cross-dimensional projection, we exploit the prior knowledge in the atlas to align the standard plane frame in a freehand scan. A semantic-aware contrastive-based approach is further proposed to align the frames that are off standard planes based on their anatomical similarity. In the experiment, we enhance the existing assessment of freehand image localization by comparing the transformation of its estimated pose towards standard plane with the corresponding probe motion, which reflects the actual view change in 3D anatomy. Extensive results on two clinical head biometry tasks show that Pose-GuideNet not only accurately predicts pose but also successfully predicts the direction of the fetal head. Evaluations with probe motions further demonstrate the feasibility of adopting Pose-GuideNet for freehand ultrasound-assisted navigation in a sensor-free environment.
Authors: Jiaming Liu, Hongyuan Liu, Zhili Qin, Wei Han, Yulu Fan, Qinli Yang, Junming Shao
Abstract: The dynamic nature of open-world scenarios has attracted more attention to class incremental learning (CIL). However, existing CIL methods typically presume the availability of complete ground-truth labels throughout the training process, an assumption rarely met in practical applications. Consequently, this paper explores a more challenging problem of unsupervised class incremental learning (UCIL). The essence of addressing this problem lies in effectively capturing comprehensive feature representations and discovering unknown novel classes. To achieve this, we first model the knowledge of class distribution by exploiting fine-grained prototypes. Subsequently, a granularity alignment technique is introduced to enhance the unsupervised class discovery. Additionally, we proposed a strategy to minimize overlap between novel and existing classes, thereby preserving historical knowledge and mitigating the phenomenon of catastrophic forgetting. Extensive experiments on the five datasets demonstrate that our approach significantly outperforms current state-of-the-art methods, indicating the effectiveness of the proposed method.
Authors: Yuncheng Jiang, Yiwen Hu, Zixun Zhang, Jun Wei, Chun-Mei Feng, Xuemei Tang, Xiang Wan, Yong Liu, Shuguang Cui, Zhen Li
Abstract: Endorectal ultrasound (ERUS) is an important imaging modality that provides high reliability for diagnosing the depth and boundary of invasion in colorectal cancer. However, the lack of a large-scale ERUS dataset with high-quality annotations hinders the development of automatic ultrasound diagnostics. In this paper, we collected and annotated the first benchmark dataset that covers diverse ERUS scenarios, i.e. colorectal cancer segmentation, detection, and infiltration depth staging. Our ERUS-10K dataset comprises 77 videos and 10,000 high-resolution annotated frames. Based on this dataset, we further introduce a benchmark model for colorectal cancer segmentation, named the Adaptive Sparse-context TRansformer (ASTR). ASTR is designed based on three considerations: scanning mode discrepancy, temporal information, and low computational complexity. For generalizing to different scanning modes, the adaptive scanning-mode augmentation is proposed to convert between raw sector images and linear scan ones. For mining temporal information, the sparse-context transformer is incorporated to integrate inter-frame local and global features. For reducing computational complexity, the sparse-context block is introduced to extract contextual features from auxiliary frames. Finally, on the benchmark dataset, the proposed ASTR model achieves a 77.6% Dice score in rectal cancer segmentation, largely outperforming previous state-of-the-art methods.
Authors: Gen Li, Nikolaos Tsagkas, Jifei Song, Ruaridh Mon-Williams, Sethu Vijayakumar, Kun Shao, Laura Sevilla-Lara
Abstract: Affordance, defined as the potential actions that an object offers, is crucial for robotic manipulation tasks. A deep understanding of affordance can lead to more intelligent AI systems. For example, such knowledge directs an agent to grasp a knife by the handle for cutting and by the blade when passing it to someone. In this paper, we present a streamlined affordance learning system that encompasses data collection, effective model training, and robot deployment. First, we collect training data from egocentric videos in an automatic manner. Different from previous methods that focus only on the object graspable affordance and represent it as coarse heatmaps, we cover both graspable (e.g., object handles) and functional affordances (e.g., knife blades, hammer heads) and extract data with precise segmentation masks. We then propose an effective model, termed Geometry-guided Affordance Transformer (GKT), to train on the collected data. GKT integrates an innovative Depth Feature Injector (DFI) to incorporate 3D shape and geometric priors, enhancing the model's understanding of affordances. To enable affordance-oriented manipulation, we further introduce Aff-Grasp, a framework that combines GKT with a grasp generation model. For comprehensive evaluation, we create an affordance evaluation dataset with pixel-wise annotations, and design real-world tasks for robot experiments. The results show that GKT surpasses the state-of-the-art by 15.9% in mIoU, and Aff-Grasp achieves high success rates of 95.5% in affordance prediction and 77.1% in successful grasping among 179 trials, including evaluations with seen, unseen objects, and cluttered scenes.
Authors: Bhavna Gopal, Huanrui Yang, Jingyang Zhang, Mark Horton, Yiran Chen
Abstract: Adversarial training enhances neural network robustness but suffers from a tendency to overfit and increased generalization errors on clean data. This work introduces CLAT, an innovative approach that mitigates adversarial overfitting by introducing parameter efficiency into the adversarial training process, improving both clean accuracy and adversarial robustness. Instead of tuning the entire model, CLAT identifies and fine-tunes robustness-critical layers - those predominantly learning non-robust features - while freezing the remaining model to enhance robustness. It employs dynamic critical layer selection to adapt to changes in layer criticality throughout the fine-tuning process. Empirically, CLAT can be applied on top of existing adversarial training methods, significantly reduces the number of trainable parameters by approximately 95%, and achieves more than a 2% improvement in adversarial robustness compared to baseline methods.
Authors: Ruoxi Chen, Haibo Jin, Haibin Zheng, Jinyin Chen, Zhenguang Liu
Abstract: The vulnerabilities of deep learning models towards adversarial attacks have attracted increasing attention, especially when models are deployed in security-critical domains. Numerous defense methods, including reactive and proactive ones, have been proposed for model robustness improvement. Reactive defenses, such as conducting transformations to remove perturbations, usually fail to handle large perturbations. The proactive defenses that involve retraining, suffer from the attack dependency and high computation cost. In this paper, we consider defense methods from the general effect of adversarial attacks that take on neurons inside the model. We introduce the concept of neuron influence, which can quantitatively measure neurons' contribution to correct classification. Then, we observe that almost all attacks fool the model by suppressing neurons with larger influence and enhancing those with smaller influence. Based on this, we propose \emph{Neuron-level Inverse Perturbation} (NIP), a novel defense against general adversarial attacks. It calculates neuron influence from benign examples and then modifies input examples by generating inverse perturbations that can in turn strengthen neurons with larger influence and weaken those with smaller influence.
Authors: Jinpeng Lin, Zhihao Liang, Shengheng Deng, Lile Cai, Tao Jiang, Tianrui Li, Kui Jia, Xun Xu
Abstract: 3D object detection has recently received much attention due to its great potential in autonomous vehicle (AV). The success of deep learning based object detectors relies on the availability of large-scale annotated datasets, which is time-consuming and expensive to compile, especially for 3D bounding box annotation. In this work, we investigate diversity-based active learning (AL) as a potential solution to alleviate the annotation burden. Given limited annotation budget, only the most informative frames and objects are automatically selected for human to annotate. Technically, we take the advantage of the multimodal information provided in an AV dataset, and propose a novel acquisition function that enforces spatial and temporal diversity in the selected samples. We benchmark the proposed method against other AL strategies under realistic annotation cost measurement, where the realistic costs for annotating a frame and a 3D bounding box are both taken into consideration. We demonstrate the effectiveness of the proposed method on the nuScenes dataset and show that it outperforms existing AL strategies significantly.
Authors: Guang Li, Ren Togo, Takahiro Ogawa, Miki Haseyama
Abstract: In this study, we propose a novel method called region-guided masked image modeling (RGMIM) for learning meaningful representations from X-ray images. Our method adopts a new masking strategy that utilizes organ mask information to identify valid regions for learning more meaningful representations. We conduct quantitative evaluations on an open lung X-ray image dataset as well as masking ratio hyperparameter studies. When using the entire training set, RGMIM outperformed other comparable methods, achieving a 0.962 lung disease detection accuracy. Specifically, RGMIM significantly improved performance in small data volumes, such as 5% and 10% of the training set compared to other methods. RGMIM can mask more valid regions, facilitating the learning of discriminative representations and the subsequent high-accuracy lung disease detection. RGMIM outperforms other state-of-the-art self-supervised learning methods in experiments, particularly when limited training data is used.
Authors: Mengmi Zhang, Giorgia Dellaferrera, Ankur Sikarwar, Caishun Chen, Marcelo Armendariz, Noga Mudrik, Prachi Agrawal, Spandan Madan, Mranmay Shetty, Andrei Barbu, Haochen Yang, Tanishq Kumar, Shui'Er Han, Aman Raj Singh, Meghna Sadwani, Stella Dellaferrera, Michele Pizzochero, Brandon Tang, Yew Soon Ong, Hanspeter Pfister, Gabriel Kreiman
Abstract: As AI algorithms increasingly participate in daily activities, it becomes critical to ascertain whether the agents we interact with are human or not. To address this question, we turn to the Turing test and systematically benchmark current AIs in their abilities to imitate humans in three language tasks (Image captioning, Word association, and Conversation) and three vision tasks (Object detection, Color estimation, and Attention prediction). The experiments involved 549 human agents plus 26 AI agents for dataset creation, and 1,126 human judges plus 10 AI judges, in 25,650 Turing-like tests. The results reveal that current AIs are not far from being able to impersonate humans in complex language and vision challenges. While human judges were often deceived, simple AI judges outperformed human judges in distinguishing human answers from AI answers. The results of imitation tests are only minimally correlated with standard performance metrics in AI. Thus, evaluating whether a machine can pass as a human constitutes an important independent test to evaluate AI algorithms. The curated, large-scale, Turing datasets introduced here and their evaluation metrics provide new benchmarks and insights to assess whether an agent is human or not and emphasize the relevance of rigorous, systematic, and quantitative imitation tests in these and other AI domains.
Authors: Jing Xu, Wentao Shi, Pan Gao, Zhengwei Wang, Qizhu Li
Abstract: In recent works on semantic segmentation, there has been a significant focus on designing and integrating transformer-based encoders. However, less attention has been given to transformer-based decoders. We emphasize that the decoder stage is equally vital as the encoder in achieving superior segmentation performance. It disentangles and refines high-level cues, enabling precise object boundary delineation at the pixel level. In this paper, we introduce a novel transformer-based decoder called MUSTER, which seamlessly integrates with hierarchical encoders and consistently delivers high-quality segmentation results, regardless of the encoder architecture. Furthermore, we present a variant of MUSTER that reduces FLOPS while maintaining performance. MUSTER incorporates carefully designed multi-head skip attention (MSKA) units and introduces innovative upsampling operations. The MSKA units enable the fusion of multi-scale features from the encoder and decoder, facilitating comprehensive information integration. The upsampling operation leverages encoder features to enhance object localization and surpasses traditional upsampling methods, improving mIoU (mean Intersection over Union) by 0.4% to 3.2%. On the challenging ADE20K dataset, our best model achieves a single-scale mIoU of 50.23 and a multi-scale mIoU of 51.88, which is on-par with the current state-of-the-art model. Remarkably, we achieve this while significantly reducing the number of FLOPs by 61.3%. Our source code and models are publicly available at: https://github.com/shiwt03/MUSTER.
Authors: Raz Lapid, Eylon Mizrahi, Moshe Sipper
Abstract: Adversarial attacks on deep-learning models have been receiving increased attention in recent years. Work in this area has mostly focused on gradient-based techniques, so-called "white-box" attacks, wherein the attacker has access to the targeted model's internal parameters; such an assumption is usually unrealistic in the real world. Some attacks additionally use the entire pixel space to fool a given model, which is neither practical nor physical (i.e., real-world). On the contrary, we propose herein a direct, black-box, gradient-free method that uses the learned image manifold of a pretrained generative adversarial network (GAN) to generate naturalistic physical adversarial patches for object detectors. To our knowledge this is the first and only method that performs black-box physical attacks directly on object-detection models, which results with a model-agnostic attack. We show that our proposed method works both digitally and physically. We compared our approach against four different black-box attacks with different configurations. Our approach outperformed all other approaches that were tested in our experiments by a large margin.
Authors: Fan-Yun Sun, Jonathan Tremblay, Valts Blukis, Kevin Lin, Danfei Xu, Boris Ivanovic, Peter Karkus, Stan Birchfield, Dieter Fox, Ruohan Zhang, Yunzhu Li, Jiajun Wu, Marco Pavone, Nick Haber
Abstract: We propose Filtering Inversion (FINV), a learning framework and optimization process that predicts a renderable 3D object representation from one or few partial views. FINV addresses the challenge of synthesizing novel views of objects from partial observations, spanning cases where the object is not entirely in view, is partially occluded, or is only observed from similar views. To achieve this, FINV learns shape priors by training a 3D generative model. At inference, given one or more views of a novel real-world object, FINV first finds a set of latent codes for the object by inverting the generative model from multiple initial seeds. Maintaining the set of latent codes, FINV filters and resamples them after receiving each new observation, akin to particle filtering. The generator is then finetuned for each latent code on the available views in order to adapt to novel objects. We show that FINV successfully synthesizes novel views of real-world objects (e.g., chairs, tables, and cars), even if the generative prior is trained only on synthetic objects. The ability to address the sim-to-real problem allows FINV to be used for object categories without real-world datasets. FINV achieves state-of-the-art performance on multiple real-world datasets, recovers object shape and texture from partial and sparse views, is robust to occlusion, and is able to incrementally improve its representation with more observations.
Authors: Jiarui Sun, Girish Chowdhary
Abstract: Stochastic Human Motion Prediction (HMP) aims to predict multiple possible future human pose sequences from observed ones. Most prior works learn motion distributions through encoding-decoding in the latent space, which does not preserve motion's spatial-temporal structure. While effective, these methods often require complex, multi-stage training and yield predictions that are inconsistent with the provided history and can be physically unrealistic. To address these issues, we propose CoMusion, a single-stage, end-to-end diffusion-based stochastic HMP framework. CoMusion is inspired from the insight that a smooth future pose initialization improves prediction performance, a strategy not previously utilized in stochastic models but evidenced in deterministic works. To generate such initialization, CoMusion's motion predictor starts with a Transformer-based network for initial reconstruction of corrupted motion. Then, a graph convolutional network (GCN) is employed to refine the prediction considering past observations in the discrete cosine transformation (DCT) space. Our method, facilitated by the Transformer-GCN module design and a proposed variance scheduler, excels in predicting accurate, realistic, and consistent motions, while maintaining appropriate diversity. Experimental results on benchmark datasets demonstrate that CoMusion surpasses prior methods across metrics, while demonstrating superior generation quality. Our Code is released at https://github.com/jsun57/CoMusion/ .
Authors: Pinxue Guo, Tony Huang, Peiyang He, Xuefeng Liu, Tianjun Xiao, Zhaoyu Chen, Wenqiang Zhang
Abstract: Open-vocabulary Video Instance Segmentation (OpenVIS) can simultaneously detect, segment, and track arbitrary object categories in a video, without being constrained to categories seen during training. In this work, we propose InstFormer, a carefully designed framework for the OpenVIS task that achieves powerful open-vocabulary capabilities through lightweight fine-tuning with limited-category data. InstFormer begins with the open-world mask proposal network, encouraged to propose all potential instance class-agnostic masks by the contrastive instance margin loss. Next, we introduce InstCLIP, adapted from pre-trained CLIP with Instance Guidance Attention, which encodes open-vocabulary instance tokens efficiently. These instance tokens not only enable open-vocabulary classification but also offer strong universal tracking capabilities. Furthermore, to prevent the tracking module from being constrained by the training data with limited categories, we propose the universal rollout association, which transforms the tracking problem into predicting the next frame's instance tracking token. The experimental results demonstrate the proposed InstFormer achieve state-of-the-art capabilities on a comprehensive OpenVIS evaluation benchmark, while also achieves competitive performance in fully supervised VIS task.
Authors: Hao Xu, Jia Liu, Yang Shen, Kenan Lou, Yanxia Bao, Ruihua Zhang, Shuyue Zhou, Hongsen Zhao, Shuai Wang
Abstract: Graph Pooling technology plays an important role in graph node classification tasks. Sorting pooling technologies maintain large-value units for pooling graphs of varying sizes. However, by analyzing the statistical characteristic of activated units after pooling, we found that a large number of units dropped by sorting pooling are negative-value units that contain useful information and can contribute considerably to the final decision. To maintain more useful information, a novel pooling technology, called Geometric Pooling (GP), was proposed to contain the unique node features with negative values by measuring the similarity of all node features. We reveal the effectiveness of GP from the entropy reduction view. The experiments were conducted on TUdatasets to show the effectiveness of GP. The results showed that the proposed GP outperforms the SOTA graph pooling technologies by 1%\sim5% with fewer parameters.
Authors: Lala Shakti Swarup Ray, Bo Zhou, Lars Krupp, Sungho Suh, Paul Lukowicz
Abstract: Accurate camera calibration is crucial for various computer vision applications. However, measuring calibration accuracy in the real world is challenging due to the lack of datasets with ground truth to evaluate them. In this paper, we present SynthCal, a synthetic camera calibration benchmarking pipeline that generates images of calibration patterns to measure and enable accurate quantification of calibration algorithm performance in camera parameter estimation. We present a SynthCal generated calibration dataset with four common patterns, two camera types, and two environments with varying view, distortion, lighting, and noise levels for both monocular and multi-camera systems. The dataset evaluates both single and multi-view calibration algorithms by measuring re-projection and root-mean-square errors for identical patterns and camera settings. Additionally, we analyze the significance of different patterns using different calibration configurations. The experimental results demonstrate the effectiveness of SynthCal in evaluating various calibration algorithms and patterns.
Authors: Yuanjing Feng, Lei Xie, Jingqiang Wang, Qiyuan Tian, Jianzhong He, Qingrun Zeng, Fei Gao
Abstract: Tractography traces the peak directions extracted from fiber orientation distribution (FOD) suffering from ambiguous spatial correspondences between diffusion directions and fiber geometry, which is prone to producing erroneous tracks while missing true positive connections. The peaks-based tractography methods 'locally' reconstructed streamlines in 'single to single' manner, thus lacking of global information about the trend of the whole fiber bundle. In this work, we propose a novel tractography method based on a bundle-specific tractogram distribution function by using a higher-order streamline differential equation, which reconstructs the streamline bundles in 'cluster to cluster' manner. A unified framework for any higher-order streamline differential equation is presented to describe the fiber bundles with disjoint streamlines defined based on the diffusion tensor vector field. At the global level, the tractography process is simplified as the estimation of bundle-specific tractogram distribution (BTD) coefficients by minimizing the energy optimization model, and is used to characterize the relations between BTD and diffusion tensor vector under the prior guidance by introducing the tractogram bundle information to provide anatomic priors. Experiments are performed on simulated Hough, Sine, Circle data, ISMRM 2015 Tractography Challenge data, FiberCup data, and in vivo data from the Human Connectome Project (HCP) data for qualitative and quantitative evaluation. The results demonstrate that our approach can reconstruct the complex global fiber bundles directly. BTD reduces the error deviation and accumulation at the local level and shows better results in reconstructing long-range, twisting, and large fanning tracts.
Authors: Zhenlong Yuan, Jiakai Cao, Zhaoqi Wang, Zhaoxin Li
Abstract: The reconstruction of textureless areas has long been a challenging problem in MVS due to lack of reliable pixel correspondences between images. In this paper, we propose the Textureless-aware Segmentation And Correlative Refinement guided Multi-View Stereo (TSAR-MVS), a novel method that effectively tackles challenges posed by textureless areas in 3D reconstruction through filtering, refinement and segmentation. First, we implement the joint hypothesis filtering, a technique that merges a confidence estimator with a disparity discontinuity detector to eliminate incorrect depth estimations. Second, to spread the pixels with confident depth, we introduce an iterative correlation refinement strategy that leverages RANSAC to generate 3D planes based on superpixels, succeeded by a weighted median filter for broadening the influence of accurately determined pixels. Finally, we present a textureless-aware segmentation method that leverages edge detection and line detection for accurately identify large textureless regions for further depth completion. Experiments on ETH3D, Tanks & Temples and Strecha datasets demonstrate the superior performance and strong generalization capability of our proposed method.
Authors: Anuj Rai, Parsheel Kumar Tiwari, Jyotishna Baishya, Ram Prakash Sharma, Somnath Dey
Abstract: Automatic fingerprint recognition systems suffer from the threat of presentation attacks due to their wide range of deployment in areas including national borders and commercial applications. A presentation attack can be performed by creating a spoof of a user's fingerprint with or without their consent. This paper presents a dynamic ensemble of deep CNN and handcrafted features to detect presentation attacks in known-material and unknown-material protocols of the liveness detection competition. The proposed presentation attack detection model, in this way, utilizes the capabilities of both deep CNN and handcrafted features techniques and exhibits better performance than their individual performances. We have validated our proposed method on benchmark databases from the Liveness Detection Competition in 2015, 2017, and 2019, yielding overall accuracy of 96.10\%, 96.49\%, and 94.99\% on them, respectively. The proposed method outperforms state-of-the-art methods in terms of classification accuracy.
Authors: Qianliang Wu, Yaqing Ding, Lei Luo, Haobo Jiang, Shuo Gu, Chuanwei Zhou, Jin Xie, Jian Yang
Abstract: Point Cloud Registration (PCR) is a critical and challenging task in computer vision. One of the primary difficulties in PCR is identifying salient and meaningful points that exhibit consistent semantic and geometric properties across different scans. Previous methods have encountered challenges with ambiguous matching due to the similarity among patch blocks throughout the entire point cloud and the lack of consideration for efficient global geometric consistency. To address these issues, we propose a new framework that includes several novel techniques. Firstly, we introduce a semantic-aware geometric encoder that combines object-level and patch-level semantic information. This encoder significantly improves registration recall by reducing ambiguity in patch-level superpoint matching. Additionally, we incorporate a prior knowledge approach that utilizes an intrinsic shape signature to identify salient points. This enables us to extract the most salient super points and meaningful dense points in the scene. Secondly, we introduce an innovative transformer that encodes High-Order (HO) geometric features. These features are crucial for identifying salient points within initial overlap regions while considering global high-order geometric consistency. To optimize this high-order transformer further, we introduce an anchor node selection strategy. By encoding inter-frame triangle or polyhedron consistency features based on these anchor nodes, we can effectively learn high-order geometric features of salient super points. These high-order features are then propagated to dense points and utilized by a Sinkhorn matching module to identify key correspondences for successful registration. In our experiments conducted on well-known datasets such as 3DMatch/3DLoMatch and KITTI, our approach has shown promising results, highlighting the effectiveness of our novel method.
Authors: Changhe Chen, Mozhgan Pourkeshavarz, Amir Rasouli
Abstract: Benchmarking is a common method for evaluating trajectory prediction models for autonomous driving. Existing benchmarks rely on datasets, which are biased towards more common scenarios, such as cruising, and distance-based metrics that are computed by averaging over all scenarios. Following such a regiment provides a little insight into the properties of the models both in terms of how well they can handle different scenarios and how admissible and diverse their outputs are. There exist a number of complementary metrics designed to measure the admissibility and diversity of trajectories, however, they suffer from biases, such as length of trajectories. In this paper, we propose a new benChmarking paRadIgm for evaluaTing trajEctoRy predIction Approaches (CRITERIA). Particularly, we propose 1) a method for extracting driving scenarios at varying levels of specificity according to the structure of the roads, models' performance, and data properties for fine-grained ranking of prediction models; 2) A set of new bias-free metrics for measuring diversity, by incorporating the characteristics of a given scenario, and admissibility, by considering the structure of roads and kinematic compliancy, motivated by real-world driving constraints. 3) Using the proposed benchmark, we conduct extensive experimentation on a representative set of the prediction models using the large scale Argoverse dataset. We show that the proposed benchmark can produce a more accurate ranking of the models and serve as a means of characterizing their behavior. We further present ablation studies to highlight contributions of different elements that are used to compute the proposed metrics.
Authors: Jiaming Zhang, Xingjun Ma, Xin Wang, Lingyu Qiu, Jiaqi Wang, Yu-Gang Jiang, Jitao Sang
Abstract: With the rapid advancement of multimodal learning, pre-trained Vision-Language Models (VLMs) such as CLIP have demonstrated remarkable capacities in bridging the gap between visual and language modalities. However, these models remain vulnerable to adversarial attacks, particularly in the image modality, presenting considerable security risks. This paper introduces Adversarial Prompt Tuning (AdvPT), a novel technique to enhance the adversarial robustness of image encoders in VLMs. AdvPT innovatively leverages learnable text prompts and aligns them with adversarial image embeddings, to address the vulnerabilities inherent in VLMs without the need for extensive parameter training or modification of the model architecture. We demonstrate that AdvPT improves resistance against white-box and black-box adversarial attacks and exhibits a synergistic effect when combined with existing image-processing-based defense techniques, further boosting defensive capabilities. Comprehensive experimental analyses provide insights into adversarial prompt tuning, a novel paradigm devoted to improving resistance to adversarial images through textual input modifications, paving the way for future robust multimodal learning research. These findings open up new possibilities for enhancing the security of VLMs. Our code is available at https://github.com/jiamingzhang94/Adversarial-Prompt-Tuning.
URLs: https://github.com/jiamingzhang94/Adversarial-Prompt-Tuning.
Authors: Yu Chi, Fangneng Zhan, Sibo Wu, Christian Theobalt, Adam Kortylewski
Abstract: Progress in 3D computer vision tasks demands a huge amount of data, yet annotating multi-view images with 3D-consistent annotations, or point clouds with part segmentation is both time-consuming and challenging. This paper introduces DatasetNeRF, a novel approach capable of generating infinite, high-quality 3D-consistent 2D annotations alongside 3D point cloud segmentations, while utilizing minimal 2D human-labeled annotations. Specifically, we leverage the strong semantic prior within a 3D generative model to train a semantic decoder, requiring only a handful of fine-grained labeled samples. Once trained, the decoder efficiently generalizes across the latent space, enabling the generation of infinite data. The generated data is applicable across various computer vision tasks, including video segmentation and 3D point cloud segmentation. Our approach not only surpasses baseline models in segmentation quality, achieving superior 3D consistency and segmentation precision on individual images, but also demonstrates versatility by being applicable to both articulated and non-articulated generative models. Furthermore, we explore applications stemming from our approach, such as 3D-aware semantic editing and 3D inversion.
Authors: Xiaoyue Mi, Fan Tang, Yepeng Weng, Danding Wang, Juan Cao, Sheng Tang, Peng Li, Yang Liu
Abstract: Despite the effectiveness in improving the robustness of neural networks, adversarial training has suffered from the natural accuracy degradation problem, i.e., accuracy on natural samples has reduced significantly. In this study, we reveal that natural accuracy degradation is highly related to the disruption of the natural sample topology in the representation space by quantitative and qualitative experiments. Based on this observation, we propose Topology-pReserving Adversarial traINing (TRAIN) to alleviate the problem by preserving the topology structure of natural samples from a standard model trained only on natural samples during adversarial training. As an additional regularization, our method can be combined with various popular adversarial training algorithms, taking advantage of both sides. Extensive experiments on CIFAR-10, CIFAR-100, and Tiny ImageNet show that our proposed method achieves consistent and significant improvements over various strong baselines in most cases. Specifically, without additional data, TRAIN achieves up to 8.86% improvement in natural accuracy and 6.33% improvement in robust accuracy.
Authors: Jiaqi Zhao, Zeyu Ding, Yong Zhou, Hancheng Zhu, Wenliang Du, Rui Yao, Abdulmotaleb El Saddik
Abstract: Object instances in remote sensing images often distribute with multi-orientations, varying scales, and dense distribution. These issues bring challenges to end-to-end oriented object detectors including multi-scale features alignment and a large number of queries. To address these limitations, we propose an end-to-end oriented detector equipped with an efficient decoder, which incorporates two technologies, Rotated RoI attention (RRoI attention) and Selective Distinct Queries (SDQ). Specifically, RRoI attention effectively focuses on oriented regions of interest through a cross-attention mechanism and aligns multi-scale features. SDQ collects queries from intermediate decoder layers and then filters similar queries to obtain distinct queries. The proposed SDQ can facilitate the optimization of one-to-one label assignment, without introducing redundant initial queries or extra auxiliary branches. Extensive experiments on five datasets demonstrate the effectiveness of our method. Notably, our method achieves state-of-the-art performance on DIOR-R (67.31% mAP), DOTA-v1.5 (67.43% mAP), and DOTA-v2.0 (53.28% mAP) with the ResNet50 backbone.
Authors: Qingxuan Lv, Yuezun Li, Junyu Dong, Sheng Chen, Hui Yu, Huiyu Zhou, Shu Zhang
Abstract: Recent DeepFake detection methods have shown excellent performance on public datasets but are significantly degraded on new forgeries. Solving this problem is important, as new forgeries emerge daily with the continuously evolving generative techniques. Many efforts have been made for this issue by seeking the commonly existing traces empirically on data level. In this paper, we rethink this problem and propose a new solution from the unsupervised domain adaptation perspective. Our solution, called DomainForensics, aims to transfer the forgery knowledge from known forgeries to new forgeries. Unlike recent efforts, our solution does not focus on data view but on learning strategies of DeepFake detectors to capture the knowledge of new forgeries through the alignment of domain discrepancies. In particular, unlike the general domain adaptation methods which consider the knowledge transfer in the semantic class category, thus having limited application, our approach captures the subtle forgery traces. We describe a new bi-directional adaptation strategy dedicated to capturing the forgery knowledge across domains. Specifically, our strategy considers both forward and backward adaptation, to transfer the forgery knowledge from the source domain to the target domain in forward adaptation and then reverse the adaptation from the target domain to the source domain in backward adaptation. In forward adaptation, we perform supervised training for the DeepFake detector in the source domain and jointly employ adversarial feature adaptation to transfer the ability to detect manipulated faces from known forgeries to new forgeries. In backward adaptation, we further improve the knowledge transfer by coupling adversarial adaptation with self-distillation on new forgeries. This enables the detector to expose new forgery features from unlabeled data and avoid forgetting the known knowledge of known...
Authors: Junyu Xie, Weidi Xie, Andrew Zisserman
Abstract: The goal of this paper is to discover, segment, and track independently moving objects in complex visual scenes. Previous approaches have explored the use of optical flow for motion segmentation, leading to imperfect predictions due to partial motion, background distraction, and object articulations and interactions. To address this issue, we introduce an appearance-based refinement method that leverages temporal consistency in video streams to correct inaccurate flow-based proposals. Our approach involves a sequence-level selection mechanism that identifies accurate flow-predicted masks as exemplars, and an object-centric architecture that refines problematic masks based on exemplar information. The model is pre-trained on synthetic data and then adapted to real-world videos in a self-supervised manner, eliminating the need for human annotations. Its performance is evaluated on multiple video segmentation benchmarks, including DAVIS, YouTubeVOS, SegTrackv2, and FBMS-59. We achieve competitive performance on single-object segmentation, while significantly outperforming existing models on the more challenging problem of multi-object segmentation. Finally, we investigate the benefits of using our model as a prompt for the per-frame Segment Anything Model.
Authors: Bruno Korbar, Yongqin Xian, Alessio Tonioni, Andrew Zisserman, Federico Tombari
Abstract: In this paper we present a text-conditioned video resampler (TCR) module that uses a pre-trained and frozen visual encoder and large language model (LLM) to process long video sequences for a task. TCR localises relevant visual features from the video given a text condition and provides them to a LLM to generate a text response. Due to its lightweight design and use of cross-attention, TCR can process more than 100 frames at a time with plain attention and without optimised implementations. We make the following contributions: (i) we design a transformer-based sampling architecture that can process long videos conditioned on a task, together with a training method that enables it to bridge pre-trained visual and language models; (ii) we identify tasks that could benefit from longer video perception; and (iii) we empirically validate its efficacy on a wide variety of evaluation tasks including NextQA, EgoSchema, and the EGO4D-LTA challenge.
Authors: Yunzhi Yan, Haotong Lin, Chenxu Zhou, Weijie Wang, Haiyang Sun, Kun Zhan, Xianpeng Lang, Xiaowei Zhou, Sida Peng
Abstract: This paper aims to tackle the problem of modeling dynamic urban streets for autonomous driving scenes. Recent methods extend NeRF by incorporating tracked vehicle poses to animate vehicles, enabling photo-realistic view synthesis of dynamic urban street scenes. However, significant limitations are their slow training and rendering speed. We introduce Street Gaussians, a new explicit scene representation that tackles these limitations. Specifically, the dynamic urban scene is represented as a set of point clouds equipped with semantic logits and 3D Gaussians, each associated with either a foreground vehicle or the background. To model the dynamics of foreground object vehicles, each object point cloud is optimized with optimizable tracked poses, along with a 4D spherical harmonics model for the dynamic appearance. The explicit representation allows easy composition of object vehicles and background, which in turn allows for scene editing operations and rendering at 135 FPS (1066 $\times$ 1600 resolution) within half an hour of training. The proposed method is evaluated on multiple challenging benchmarks, including KITTI and Waymo Open datasets. Experiments show that the proposed method consistently outperforms state-of-the-art methods across all datasets. The code will be released to ensure reproducibility.
Authors: Fengfan Zhou, Qianyu Zhou, Bangjie Yin, Hui Zheng, Xuequan Lu, Lizhuang Ma, Hefei Ling
Abstract: Face Recognition (FR) systems can be easily deceived by adversarial examples that manipulate benign face images through imperceptible perturbations. Adversarial attacks on FR encompass two types: impersonation (targeted) attacks and dodging (untargeted) attacks. Previous methods often achieve a successful impersonation attack on FR, however, it does not necessarily guarantee a successful dodging attack on FR in the black-box setting. In this paper, our key insight is that the generation of adversarial examples should perform both impersonation and dodging attacks simultaneously. To this end, we propose a novel attack method termed as Adversarial Pruning (Adv-Pruning), to fine-tune existing adversarial examples to enhance their dodging capabilities while preserving their impersonation capabilities. Adv-Pruning consists of Priming, Pruning, and Restoration stages. Concretely, we propose Adversarial Priority Quantification to measure the region-wise priority of original adversarial perturbations, identifying and releasing those with minimal impact on absolute model output variances. Then, Biased Gradient Adaptation is presented to adapt the adversarial examples to traverse the decision boundaries of both the attacker and victim by adding perturbations favoring dodging attacks on the vacated regions, preserving the prioritized features of the original perturbations while boosting dodging performance. As a result, we can maintain the impersonation capabilities of original adversarial examples while effectively enhancing dodging capabilities. Comprehensive experiments demonstrate the superiority of our method compared with state-of-the-art adversarial attack methods.
Authors: Zhe Li, Jinglin Zhao, Zheng Wang, Bocheng Ren, Debin Liu, Ziyang Zhang, Laurence T. Yang
Abstract: Masked autoencoding and generative pretraining have achieved remarkable success in computer vision and natural language processing, and more recently, they have been extended to the point cloud domain. Nevertheless, existing point cloud models suffer from the issue of information leakage due to the pre-sampling of center points, which leads to trivial proxy tasks for the models. These approaches primarily focus on local feature reconstruction, limiting their ability to capture global patterns within point clouds. In this paper, we argue that the reduced difficulty of pretext tasks hampers the model's capacity to learn expressive representations. To address these limitations, we introduce a novel solution called the Differentiable Center Sampling Network (DCS-Net). It tackles the information leakage problem by incorporating both global feature reconstruction and local feature reconstruction as non-trivial proxy tasks, enabling simultaneous learning of both the global and local patterns within point cloud. Experimental results demonstrate that our method enhances the expressive capacity of existing point cloud models and effectively addresses the issue of information leakage.
Authors: Youngsik Yun, Jihie Kim
Abstract: Image Captioning generates descriptive sentences from images using Vision-Language Pre-trained models (VLPs) such as BLIP, which has improved greatly. However, current methods lack the generation of detailed descriptive captions for the cultural elements depicted in the images, such as the traditional clothing worn by people from Asian cultural groups. In this paper, we propose a new framework, Culturally-aware Image Captioning (CIC), that generates captions and describes cultural elements extracted from cultural visual elements in images representing cultures. Inspired by methods combining visual modality and Large Language Models (LLMs) through appropriate prompts, our framework (1) generates questions based on cultural categories from images, (2) extracts cultural visual elements from Visual Question Answering (VQA) using generated questions, and (3) generates culturally-aware captions using LLMs with the prompts. Our human evaluation conducted on 45 participants from 4 different cultural groups with a high understanding of the corresponding culture shows that our proposed framework generates more culturally descriptive captions when compared to the image captioning baseline based on VLPs. Resources can be found at https://shane3606.github.io/cic..
Authors: Mehdi Azarafza, Mojtaba Nayyeri, Charles Steinmetz, Steffen Staab, Achim Rettberg
Abstract: Large Language Models (LLMs) have garnered significant attention for their ability to understand text and images, generate human-like text, and perform complex reasoning tasks. However, their ability to generalize this advanced reasoning with a combination of natural language text for decision-making in dynamic situations requires further exploration. In this study, we investigate how well LLMs can adapt and apply a combination of arithmetic and common-sense reasoning, particularly in autonomous driving scenarios. We hypothesize that LLMs hybrid reasoning abilities can improve autonomous driving by enabling them to analyze detected object and sensor data, understand driving regulations and physical laws, and offer additional context. This addresses complex scenarios, like decisions in low visibility (due to weather conditions), where traditional methods might fall short. We evaluated Large Language Models (LLMs) based on accuracy by comparing their answers with human-generated ground truth inside CARLA. The results showed that when a combination of images (detected objects) and sensor data is fed into the LLM, it can offer precise information for brake and throttle control in autonomous vehicles across various weather conditions. This formulation and answers can assist in decision-making for auto-pilot systems.
Authors: Lianghui Zhu, Junwei Zhou, Yan Liu, Xin Hao, Wenyu Liu, Xinggang Wang
Abstract: Weakly supervised visual recognition using inexact supervision is a critical yet challenging learning problem. It significantly reduces human labeling costs and traditionally relies on multi-instance learning and pseudo-labeling. This paper introduces WeakSAM and solves the weakly-supervised object detection (WSOD) and segmentation by utilizing the pre-learned world knowledge contained in a vision foundation model, i.e., the Segment Anything Model (SAM). WeakSAM addresses two critical limitations in traditional WSOD retraining, i.e., pseudo ground truth (PGT) incompleteness and noisy PGT instances, through adaptive PGT generation and Region of Interest (RoI) drop regularization. It also addresses the SAM's problems of requiring prompts and category unawareness for automatic object detection and segmentation. Our results indicate that WeakSAM significantly surpasses previous state-of-the-art methods in WSOD and WSIS benchmarks with large margins, i.e. average improvements of 7.4% and 8.5%, respectively. The code is available at \url{https://github.com/hustvl/WeakSAM}.
Authors: Xiangwei Meng
Abstract: The parametric constant false alarm rate (CFAR) detection algorithms which are based on various statistical distributions, such as Gaussian, Gamma, Weibull, log-normal, G0 distribution, alpha-stable distribution, etc, are most widely used to detect the ship targets in SAR image at present. However, the clutter background in SAR images is complicated and variable. When the actual clutter background deviates from the assumed statistical distribution, the performance of the parametric CFAR detector will deteriorate. In addition to the parametric CFAR schemes, there is another class of nonparametric CFAR detectors which can maintain a constant false alarm rate for the target detection without the assumption of a known clutter distribution. In this work, the Wilcoxon nonparametric CFAR scheme for ship detection in SAR image is proposed and analyzed, and a closed form of the false alarm rate for the Wilcoxon nonparametric detector to determine the decision threshold is presented. By comparison with several typical parametric CFAR schemes on Radarsat-2, ICEYE-X6 and Gaofen-3 SAR images, the robustness of the Wilcoxon nonparametric detector to maintain a good false alarm performance in different detection backgrounds is revealed, and its detection performance for the weak ship in rough sea surface is improved to some extent. Moreover, the Wilcoxon nonparametric detector can suppress the false alarms resulting from the sidelobes at some degree and its detection speed is fast.
Authors: Dovile Juodelyte, Yucheng Lu, Amelia Jim\'enez-S\'anchez, Sabrina Bottazzi, Enzo Ferrante, Veronika Cheplygina
Abstract: Transfer learning has become an essential part of medical imaging classification algorithms, often leveraging ImageNet weights. The domain shift from natural to medical images has prompted alternatives such as RadImageNet, often showing comparable classification performance. However, it remains unclear whether the performance gains from transfer learning stem from improved generalization or shortcut learning. To address this, we conceptualize confounders by introducing the Medical Imaging Contextualized Confounder Taxonomy (MICCAT) and investigate a range of confounders across it -- whether synthetic or sampled from the data -- using two public chest X-ray and CT datasets. We show that ImageNet and RadImageNet achieve comparable classification performance, yet ImageNet is much more prone to overfitting to confounders. We recommend that researchers using ImageNet-pretrained models reexamine their model robustness by conducting similar experiments. Our code and experiments are available at https://github.com/DovileDo/source-matters.
Authors: Junhui Yin, Xinyu Zhang, Lin Wu, Xiaojie Wang
Abstract: Current pre-trained vision-language models, such as CLIP, have demonstrated remarkable zero-shot generalization capabilities across various downstream tasks. However, their performance significantly degrades when test inputs exhibit different distributions. In this paper, we explore the concept of test-time prompt tuning (TTPT), which facilitates the adaptation of the CLIP model to novel downstream tasks through a one-step unsupervised optimization that involves only test samples. Inspired by in-context learning in natural language processing (NLP), we propose In-Context Prompt Learning (InCPL) for test-time visual recognition tasks, which empowers a pre-trained vision-language model with labeled examples as context information on downstream task. Specifically, InCPL associates a new test sample with very few labeled examples (sometimes just one) as context information, enabling reliable label estimation for the test sample and facilitating model adaptation. To achieve this, InCPL employs an efficient language-to-vision translator to explore the textual prior information for visual prompt learning. Further, we introduce a context-aware unsupervised loss to optimize visual prompts tailored to test samples. Finally, we design a cyclic learning strategy for visual and textual prompts to ensure mutual synergy across different modalities. This enables a pre-trained, frozen CLIP model to adapt to any task using its learned adaptive prompt. Our method demonstrates superior performance and achieves state-of-the-art results across various downstream datasets.
Authors: Zikun Xu, Jianqiang Wang, Shaobing Xu
Abstract: LiDAR-based 3D occupancy prediction evolved rapidly alongside the emergence of large datasets. Nevertheless, the potential of existing diverse datasets remains underutilized as they kick in individually. Models trained on a specific dataset often suffer considerable performance degradation when deployed to real-world scenarios or datasets involving disparate LiDARs. This paper aims to develop a generalized model called MergeOcc, to simultaneously handle different LiDARs by leveraging multiple datasets. The gaps among LiDAR datasets primarily manifest in geometric disparities and semantic inconsistencies. Thus, MergeOcc incorporates a novel model featuring a geometric realignment module and a semantic label mapping module to enable multiple datasets training (MDT). The effectiveness of MergeOcc is validated through experiments on two prominent datasets for autonomous vehicles: OpenOccupancy-nuScenes and SemanticKITTI. The results demonstrate its enhanced robustness and remarkable performance across both types of LiDARs, outperforming several SOTA multi-modality methods. Notably, despite using an identical model architecture and hyper-parameter set, MergeOcc can significantly surpass the baseline due to its exposure to more diverse data. MergeOcc is considered the first cross-dataset 3D occupancy prediction pipeline that effectively bridges the domain gap for seamless deployment across heterogeneous platforms.
Authors: Yew Ken Chia, Vernon Toh Yan Han, Deepanway Ghosal, Lidong Bing, Soujanya Poria
Abstract: Large multimodal models extend the impressive capabilities of large language models by integrating multimodal understanding abilities. However, it is not clear how they can emulate the general intelligence and reasoning ability of humans. As recognizing patterns and abstracting concepts are key to general intelligence, we introduce PuzzleVQA, a collection of 2000 puzzle instances based on abstract patterns. With this dataset, we evaluate large multimodal models with abstract patterns based on fundamental concepts, including colors, numbers, sizes, and shapes. Through our experiments on state-of-the-art large multimodal models, we find that they are not able to generalize well to simple abstract patterns. Notably, GPT-4V achieves a score of 46.4% on single-concept puzzles, which shows that state-of-the-art models struggle on our dataset. To diagnose the reasoning challenges in large multimodal models, we progressively guide the models with our ground truth reasoning explanations for visual perception, inductive reasoning, and deductive reasoning. Our systematic analysis finds that the main bottlenecks of GPT-4V are weaker visual perception and inductive reasoning abilities. Through this work, we hope to shed light on the limitations of large multimodal models and how they can better emulate human cognitive processes in the future. Our data and code are available at https://puzzlevqa.github.io
Authors: Jiaqi Yue, Jiancheng Zhao, Chunhui Zhao
Abstract: Generalized zero-shot learning (GZSL) focuses on recognizing seen and unseen classes against domain shift problem (DSP) where data of unseen classes may be misclassified as seen classes. However, existing GZSL is still limited to seen domains. In the current work, we pioneer cross-domain GZSL (CDGZSL) which addresses GZSL towards unseen domains. Different from existing GZSL methods which alleviate DSP by generating features of unseen classes with semantics, CDGZSL needs to construct a common feature space across domains and acquire the corresponding intrinsic semantics shared among domains to transfer from seen to unseen domains. Considering the information asymmetry problem caused by redundant class semantics annotated with large language models (LLMs), we present Meta Domain Alignment Semantic Refinement (MDASR). Technically, MDASR consists of two parts: Inter-class Similarity Alignment (ISA), which eliminates the non-intrinsic semantics not shared across all domains under the guidance of inter-class feature relationships, and Unseen-class Meta Generation (UMG), which preserves intrinsic semantics to maintain connectivity between seen and unseen classes by simulating feature generation. MDASR effectively aligns the redundant semantic space with the common feature space, mitigating the information asymmetry in CDGZSL. The effectiveness of MDASR is demonstrated on the Office-Home and Mini-DomainNet, and we have shared the LLM-based semantics for these datasets as the benchmark.
Authors: Sukhbinder Singh, Saeed S. Jahromi, Roman Orus
Abstract: Convolutional neural networks (CNNs) are one of the most widely used neural network architectures, showcasing state-of-the-art performance in computer vision tasks. Although larger CNNs generally exhibit higher accuracy, their size can be effectively reduced by ``tensorization'' while maintaining accuracy, namely, replacing the convolution kernels with compact decompositions such as Tucker, Canonical Polyadic decompositions, or quantum-inspired decompositions such as matrix product states, and directly training the factors in the decompositions to bias the learning towards low-rank decompositions. But why doesn't tensorization seem to impact the accuracy adversely? We explore this by assessing how \textit{truncating} the convolution kernels of \textit{dense} (untensorized) CNNs impact their accuracy. Specifically, we truncated the kernels of (i) a vanilla four-layer CNN and (ii) ResNet-50 pre-trained for image classification on CIFAR-10 and CIFAR-100 datasets. We found that kernels (especially those inside deeper layers) could often be truncated along several cuts resulting in significant loss in kernel norm but not in classification accuracy. This suggests that such ``correlation compression'' (underlying tensorization) is an intrinsic feature of how information is encoded in dense CNNs. We also found that aggressively truncated models could often recover the pre-truncation accuracy after only a few epochs of re-training, suggesting that compressing the internal correlations of convolution layers does not often transport the model to a worse minimum. Our results can be applied to tensorize and compress CNN models more effectively.
Authors: Renrui Zhang, Dongzhi Jiang, Yichi Zhang, Haokun Lin, Ziyu Guo, Pengshuo Qiu, Aojun Zhou, Pan Lu, Kai-Wei Chang, Peng Gao, Hongsheng Li
Abstract: The remarkable progress of Multi-modal Large Language Models (MLLMs) has garnered unparalleled attention, due to their superior performance in visual contexts. However, their capabilities in visual math problem-solving remain insufficiently evaluated and understood. We investigate current benchmarks to incorporate excessive visual content within textual questions, which potentially assist MLLMs in deducing answers without truly interpreting the input diagrams. To this end, we introduce MathVerse, an all-around visual math benchmark designed for an equitable and in-depth evaluation of MLLMs. We meticulously collect 2,612 high-quality, multi-subject math problems with diagrams from publicly available sources. Each problem is then transformed by human annotators into six distinct versions, each offering varying degrees of information content in multi-modality, contributing to 15K test samples in total. This approach allows MathVerse to comprehensively assess whether and how much MLLMs can truly understand the visual diagrams for mathematical reasoning. In addition, we propose a Chain-of-Thought (CoT) evaluation strategy for a fine-grained assessment of the output answers. Rather than naively judging True or False, we employ GPT-4(V) to adaptively extract crucial reasoning steps, and then score each step with detailed error analysis, which can reveal the intermediate CoT reasoning quality by MLLMs. We hope the MathVerse benchmark may provide unique insights to guide the future development of MLLMs. Project page: https://mathverse-cuhk.github.io
Authors: Yunpeng Luo, Junlong Du, Ke Yan, Shouhong Ding
Abstract: The evolution of Diffusion Models has dramatically improved image generation quality, making it increasingly difficult to differentiate between real and generated images. This development, while impressive, also raises significant privacy and security concerns. In response to this, we propose a novel Latent REconstruction error guided feature REfinement method (LaRE^2) for detecting the diffusion-generated images. We come up with the Latent Reconstruction Error (LaRE), the first reconstruction-error based feature in the latent space for generated image detection. LaRE surpasses existing methods in terms of feature extraction efficiency while preserving crucial cues required to differentiate between the real and the fake. To exploit LaRE, we propose an Error-Guided feature REfinement module (EGRE), which can refine the image feature guided by LaRE to enhance the discriminativeness of the feature. Our EGRE utilizes an align-then-refine mechanism, which effectively refines the image feature for generated-image detection from both spatial and channel perspectives. Extensive experiments on the large-scale GenImage benchmark demonstrate the superiority of our LaRE^2, which surpasses the best SoTA method by up to 11.9%/12.1% average ACC/AP across 8 different image generators. LaRE also surpasses existing methods in terms of feature extraction cost, delivering an impressive speed enhancement of 8 times.
Authors: Songpan Wang, Xu Li, Tianxiang Jiang, Yuanlun Xie
Abstract: Facial expression recognition (FER) is vital for human-computer interaction and emotion analysis, yet recognizing expressions in low-resolution images remains challenging. This paper introduces a practical method called Dynamic Resolution Guidance for Facial Expression Recognition (DRGFER) to effectively recognize facial expressions in images with varying resolutions without compromising FER model accuracy. Our framework comprises two main components: the Resolution Recognition Network (RRN) and the Multi-Resolution Adaptation Facial Expression Recognition Network (MRAFER). The RRN determines image resolution, outputs a binary vector, and the MRAFER assigns images to suitable facial expression recognition networks based on resolution. We evaluated DRGFER on widely-used datasets RAFDB and FERPlus, demonstrating that our method retains optimal model performance at each resolution and outperforms alternative resolution approaches. The proposed framework exhibits robustness against resolution variations and facial expressions, offering a promising solution for real-world applications.
Authors: Chunxu Liu, Guozhen Zhang, Rui Zhao, Limin Wang
Abstract: Large motion poses a critical challenge in Video Frame Interpolation (VFI) task. Existing methods are often constrained by limited receptive fields, resulting in sub-optimal performance when handling scenarios with large motion. In this paper, we introduce a new pipeline for VFI, which can effectively integrate global-level information to alleviate issues associated with large motion. Specifically, we first estimate a pair of initial intermediate flows using a high-resolution feature map for extracting local details. Then, we incorporate a sparse global matching branch to compensate for flow estimation, which consists of identifying flaws in initial flows and generating sparse flow compensation with a global receptive field. Finally, we adaptively merge the initial flow estimation with global flow compensation, yielding a more accurate intermediate flow. To evaluate the effectiveness of our method in handling large motion, we carefully curate a more challenging subset from commonly used benchmarks. Our method demonstrates the state-of-the-art performance on these VFI subsets with large motion.
Authors: Runtao Liu, Ashkan Khakzar, Jindong Gu, Qifeng Chen, Philip Torr, Fabio Pizzati
Abstract: With the ability to generate high-quality images, text-to-image (T2I) models can be exploited for creating inappropriate content. To prevent misuse, existing safety measures are either based on text blacklists, which can be easily circumvented, or harmful content classification, requiring large datasets for training and offering low flexibility. Hence, we propose Latent Guard, a framework designed to improve safety measures in text-to-image generation. Inspired by blacklist-based approaches, Latent Guard learns a latent space on top of the T2I model's text encoder, where it is possible to check the presence of harmful concepts in the input text embeddings. Our proposed framework is composed of a data generation pipeline specific to the task using large language models, ad-hoc architectural components, and a contrastive learning strategy to benefit from the generated data. The effectiveness of our method is verified on three datasets and against four baselines. Code and data will be shared at https://latentguard.github.io/.
Authors: Jun Wang, Yufei Cui, Yu Mao, Nan Guan, Chun Jason Xue
Abstract: Pre-processing whole slide images (WSIs) can impact classification performance. Our study shows that using fixed hyper-parameters for pre-processing out-of-domain WSIs can significantly degrade performance. Therefore, it is critical to search domain-specific hyper-parameters during inference. However, searching for an optimal parameter set is time-consuming. To overcome this, we propose SSAPT, a novel Similarity-based Simulated Annealing approach for fast parameter tuning to enhance inference performance on out-of-domain data. The proposed SSAPT achieves 5\% to 50\% improvement in accuracy with $\times5$ times faster parameter searching speed on average.
Authors: Md Adnan Arefeen, Biplob Debnath, Md Yusuf Sarwar Uddin, Srimat Chakradhar
Abstract: Retrieval-augmented generation (RAG) systems combine the strengths of language generation and information retrieval to power many real-world applications like chatbots. Use of RAG for understanding of videos is appealing but there are two critical limitations. One-time, upfront conversion of all content in large corpus of videos into text descriptions entails high processing times. Also, not all information in the rich video data is typically captured in the text descriptions. Since user queries are not known apriori, developing a system for video to text conversion and interactive querying of video data is challenging. To address these limitations, we propose an incremental RAG system called iRAG, which augments RAG with a novel incremental workflow to enable interactive querying of a large corpus of videos. Unlike traditional RAG, iRAG quickly indexes large repositories of videos, and in the incremental workflow, it uses the index to opportunistically extract more details from select portions of the videos to retrieve context relevant to an interactive user query. Such an incremental workflow avoids long video to text conversion times, and overcomes information loss issues due to conversion of video to text, by doing on-demand query-specific extraction of details in video data. This ensures high quality of responses to interactive user queries that are often not known apriori. To the best of our knowledge, iRAG is the first system to augment RAG with an incremental workflow to support efficient interactive querying of a large corpus of videos. Experimental results on real-world datasets demonstrate 23x to 25x faster video to text ingestion, while ensuring that latency and quality of responses to interactive user queries is comparable to responses from a traditional RAG where all video data is converted to text upfront before any user querying.
Authors: Jiawei Du, Jia Guo, Weihang Zhang, Shengzhu Yang, Hanruo Liu, Huiqi Li, Ningli Wang
Abstract: The Vision-Language Foundation model is increasingly investigated in the fields of computer vision and natural language processing, yet its exploration in ophthalmology and broader medical applications remains limited. The challenge is the lack of labeled data for the training of foundation model. To handle this issue, a CLIP-style retinal image foundation model is developed in this paper. Our foundation model, RET-CLIP, is specifically trained on a dataset of 193,865 patients to extract general features of color fundus photographs (CFPs), employing a tripartite optimization strategy to focus on left eye, right eye, and patient level to reflect real-world clinical scenarios. Extensive experiments demonstrate that RET-CLIP outperforms existing benchmarks across eight diverse datasets spanning four critical diagnostic categories: diabetic retinopathy, glaucoma, multiple disease diagnosis, and multi-label classification of multiple diseases, which demonstrate the performance and generality of our foundation model. The sourse code and pre-trained model are available at https://github.com/sStonemason/RET-CLIP.
Authors: Jiaze Wang, Yi Wang, Ziyu Guo, Renrui Zhang, Donghao Zhou, Guangyong Chen, Anfeng Liu, Pheng-Ann Heng
Abstract: We introduce MM-Mixing, a multi-modal mixing alignment framework for 3D understanding. MM-Mixing applies mixing-based methods to multi-modal data, preserving and optimizing cross-modal connections while enhancing diversity and improving alignment across modalities. Our proposed two-stage training pipeline combines feature-level and input-level mixing to optimize the 3D encoder. The first stage employs feature-level mixing with contrastive learning to align 3D features with their corresponding modalities. The second stage incorporates both feature-level and input-level mixing, introducing mixed point cloud inputs to further refine 3D feature representations. MM-Mixing enhances intermodality relationships, promotes generalization, and ensures feature consistency while providing diverse and realistic training samples. We demonstrate that MM-Mixing significantly improves baseline performance across various learning scenarios, including zero-shot 3D classification, linear probing 3D classification, and cross-modal 3D shape retrieval. Notably, we improved the zero-shot classification accuracy on ScanObjectNN from 51.3% to 61.9%, and on Objaverse-LVIS from 46.8% to 51.4%. Our findings highlight the potential of multi-modal mixing-based alignment to significantly advance 3D object recognition and understanding while remaining straightforward to implement and integrate into existing frameworks.
Authors: Honghao Fu, Yufei Wang, Wenhan Yang, Bihan Wen
Abstract: Blind image quality assessment (IQA) in the wild, which assesses the quality of images with complex authentic distortions and no reference images, presents significant challenges. Given the difficulty in collecting large-scale training data, leveraging limited data to develop a model with strong generalization remains an open problem. Motivated by the robust image perception capabilities of pre-trained text-to-image (T2I) diffusion models, we propose a novel IQA method, diffusion priors-based IQA (DP-IQA), to utilize the T2I model's prior for improved performance and generalization ability. Specifically, we utilize pre-trained Stable Diffusion as the backbone, extracting multi-level features from the denoising U-Net guided by prompt embeddings through a tunable text adapter. Simultaneously, an image adapter compensates for information loss introduced by the lossy pre-trained encoder. Unlike T2I models that require full image distribution modeling, our approach targets image quality assessment, which inherently requires fewer parameters. To improve applicability, we distill the knowledge into a lightweight CNN-based student model, significantly reducing parameters while maintaining or even enhancing generalization performance. Experimental results demonstrate that DP-IQA achieves state-of-the-art performance on various in-the-wild datasets, highlighting the superior generalization capability of T2I priors in blind IQA tasks. To our knowledge, DP-IQA is the first method to apply pre-trained diffusion priors in blind IQA. Codes and checkpoints are available at https://github.com/RomGai/DP-IQA.
Authors: Wei Ye, Xinan He, Feng Ding
Abstract: In this paper, we propose a novel method for detecting DeepFakes, enhancing the generalization of detection through semantic decoupling. There are now multiple DeepFake forgery technologies that not only possess unique forgery semantics but may also share common forgery semantics. The unique forgery semantics and irrelevant content semantics may promote over-fitting and hamper generalization for DeepFake detectors. For our proposed method, after decoupling, the common forgery semantics could be extracted from DeepFakes, and subsequently be employed for developing the generalizability of DeepFake detectors. Also, to pursue additional generalizability, we designed an adaptive high-pass module and a two-stage training strategy to improve the independence of decoupled semantics. Evaluation on FF++, Celeb-DF, DFD, and DFDC datasets showcases our method's excellent detection and generalization performance. Code is available at: https://github.com/leaffeall/DFS-GDD.
Authors: Yunlong Zhang, Zhongyi Shui, Yunxuan Sun, Honglin Li, Jingxiong Li, Chenglu Zhu, Lin Yang
Abstract: Multiple Instance Learning (MIL) has demonstrated effectiveness in analyzing whole slide images (WSIs), yet it often encounters overfitting challenges in real-world applications, particularly in the form of attention over-concentration. While existing methods to alleviate this issue introduce complex modules or processing steps, such as multiple-stage training and teacher-student distillation, this paper proposes a simple yet effective regularization: Attention Entropy Maximization (AEM). Motivated by our investigation revealing a positive correlation between attention entropy and model performance, AEM incorporates a negative entropy loss for attention values into the standard MIL framework, penalizing overly concentrated attention and encouraging the model to consider a broader range of informative regions in WSIs, potentially improving its generalization capabilities. Compared to existing overfitting mitigation methods, our AEM approach offers advantages of simplicity, efficiency, and versatility. It requires no additional modules or processing steps, involves only one hyperparameter, and demonstrates compatibility with MIL frameworks and techniques. These advantages make AEM particularly attractive for practical applications. We evaluate AEM on three benchmark datasets, demonstrating consistent performance improvements over existing methods. Furthermore, AEM shows high versatility, integrating effectively with four feature extractors, two advanced MIL frameworks, three attention mechanisms, and Subsampling augmentation technique. The source code is available at \url{https://github.com/dazhangyu123/AEM}.
Authors: Shilei Cao, Yan Liu, Juepeng Zheng, Weijia Li, Runmin Dong, Haohuan Fu
Abstract: Real-world application models are commonly deployed in dynamic environments, where the target domain distribution undergoes temporal changes. Continual Test-Time Adaptation (CTTA) has recently emerged as a promising technique to gradually adapt a source-trained model to continually changing target domains. Despite recent advancements in addressing CTTA, two critical issues remain: 1) Fixed thresholds for pseudo-labeling in existing methodologies generate low-quality pseudo-labels, as model confidence varies across categories and domains; 2) Stochastic parameter restoration methods for mitigating catastrophic forgetting fail to effectively preserve critical information due to their intrinsic randomness. To tackle these challenges for detection models in CTTA scenarios, we present CTAOD, featuring three core components. Firstly, the object-level contrastive learning module extracts object-level features for contrastive learning to refine the feature representation in the target domain. Secondly, the adaptive monitoring module dynamically skips unnecessary adaptation and updates the category-specific threshold based on predicted confidence scores to enable efficiency and improve the quality of pseudo-labels. Lastly, the data-driven stochastic restoration mechanism selectively reset inactive parameters with higher possibilities, ensuring the retention of essential knowledge. We demonstrate the effectiveness of CTAOD on four CTTA object detection tasks, where CTAOD outperforms existing methods, especially achieving a 3.2 mAP improvement and a 20% increase in efficiency on the Cityscapes-to-Cityscapes-C CTTA task. The code will be released.
Authors: Tom Fischer, Yaoyao Liu, Artur Jesslen, Noor Ahmed, Prakhar Kaushik, Angtian Wang, Alan Yuille, Adam Kortylewski, Eddy Ilg
Abstract: Different from human nature, it is still common practice today for vision tasks to train deep learning models only initially and on fixed datasets. A variety of approaches have recently addressed handling continual data streams. However, extending these methods to manage out-of-distribution (OOD) scenarios has not effectively been investigated. On the other hand, it has recently been shown that non-continual neural mesh models exhibit strong performance in generalizing to such OOD scenarios. To leverage this decisive property in a continual learning setting, we propose incremental neural mesh models that can be extended with new meshes over time. In addition, we present a latent space initialization strategy that enables us to allocate feature space for future unseen classes in advance and a positional regularization term that forces the features of the different classes to consistently stay in respective latent space regions. We demonstrate the effectiveness of our method through extensive experiments on the Pascal3D and ObjectNet3D datasets and show that our approach outperforms the baselines for classification by $2-6\%$ in the in-domain and by $6-50\%$ in the OOD setting. Our work also presents the first incremental learning approach for pose estimation. Our code and model can be found at https://github.com/Fischer-Tom/iNeMo.
Authors: Mufeng Yao, Jinlong Peng, Qingdong He, Bo Peng, Hao Chen, Mingmin Chi, Chao Liu, Jon Atli Benediktsson
Abstract: Multiple object tracking (MOT) from unmanned aerial vehicle (UAV) platforms requires efficient motion modeling. This is because UAV-MOT faces both local object motion and global camera motion. Motion blur also increases the difficulty of detecting large moving objects. Previous UAV motion modeling approaches either focus only on local motion or ignore motion blurring effects, thus limiting their tracking performance and speed. To address these issues, we propose the Motion Mamba Module, which explores both local and global motion features through cross-correlation and bi-directional Mamba Modules for better motion modeling. To address the detection difficulties caused by motion blur, we also design motion margin loss to effectively improve the detection accuracy of motion blurred objects. Based on the Motion Mamba module and motion margin loss, our proposed MM-Tracker surpasses the state-of-the-art in two widely open-source UAV-MOT datasets. Code will be available.
Authors: Yike Zhang, Eduardo Davalos, Dingjie Su, Ange Lou, Jack H. Noble
Abstract: Cochlear Implant (CI) procedures involve inserting an array of electrodes into the cochlea located inside the inner ear. Mastoidectomy is a surgical procedure that uses a high-speed drill to remove part of the mastoid region of the temporal bone, providing safe access to the cochlea through the middle and inner ear. We aim to develop an intraoperative navigation system that registers plans created using 3D preoperative Computerized Tomography (CT) volumes with the 2D surgical microscope view. Herein, we propose a method to synthesize the mastoidectomy volume using only the preoperative CT scan, where the mastoid is intact. We introduce an unsupervised learning framework designed to synthesize mastoidectomy. For model training purposes, this method uses postoperative CT scans to avoid manual data cleaning or labeling, even when the region removed during mastoidectomy is visible but affected by metal artifacts, low signal-to-noise ratio, or electrode wiring. Our approach estimates mastoidectomy regions with a mean dice score of 70.0%. This approach represents a major step forward for CI intraoperative navigation by predicting realistic mastoidectomy-removed regions in preoperative planning that can be used to register the pre-surgery plan to intraoperative microscopy.
Authors: Xi Shi, Lingli Chen, Peng Wei, Xi Wu, Tian Jiang, Yonggang Luo, Lecheng Xie
Abstract: Existing Gaussian splatting methods often fall short in achieving satisfactory novel view synthesis in driving scenes, primarily due to the absence of crafty designs and geometric constraints for the involved elements. This paper introduces a novel neural rendering method termed Decoupled Hybrid Gaussian Splatting (DHGS), targeting at promoting the rendering quality of novel view synthesis for static driving scenes. The novelty of this work lies in the decoupled and hybrid pixel-level blender for road and non-road layers, without the conventional unified differentiable rendering logic for the entire scene. Still, consistency and continuity in superimposition are preserved through the proposed depth-ordered hybrid rendering strategy. Additionally, an implicit road representation comprised of a Signed Distance Function (SDF) is trained to supervise the road surface with subtle geometric attributes. Accompanied by the use of auxiliary transmittance loss and consistency loss, novel images with imperceptible boundary and elevated fidelity are ultimately obtained. Substantial experiments on the Waymo dataset prove that DHGS outperforms the state-of-the-art methods. The project page where more video evidences are given is: https://ironbrotherstyle.github.io/dhgs_web.
Authors: Wei Wang, Qing Li
Abstract: Computer vision (CV) is one of the most crucial fields in artificial intelligence. In recent years, a variety of deep learning models based on convolutional neural networks (CNNs) and Transformers have been designed to tackle diverse problems in CV. These algorithms have found practical applications in areas such as robotics and facial recognition. Despite the increasing power of current CV models, several fundamental questions remain unresolved: Why do CNNs require deep layers? What ensures the generalization ability of CNNs? Why do residual-based networks outperform fully convolutional networks like VGG? What is the fundamental difference between residual-based CNNs and Transformer-based networks? Why can CNNs utilize LoRA and pruning techniques? The root cause of these questions lies in the lack of a robust theoretical foundation for deep learning models in CV. To address these critical issues and techniques, we employ the Universal Approximation Theorem (UAT) to provide a theoretical basis for convolution- and Transformer-based models in CV. By doing so, we aim to elucidate these questions from a theoretical perspective.
Authors: Juhan Cha, Minseok Joo, Jihwan Park, Sanghyeok Lee, Injae Kim, Hyunwoo J. Kim
Abstract: Recent advancements in 3D object detection have benefited from multi-modal information from the multi-view cameras and LiDAR sensors. However, the inherent disparities between the modalities pose substantial challenges. We observe that existing multi-modal 3D object detection methods heavily rely on the LiDAR sensor, treating the camera as an auxiliary modality for augmenting semantic details. This often leads to not only underutilization of camera data but also significant performance degradation in scenarios where LiDAR data is unavailable. Additionally, existing fusion methods overlook the detrimental impact of sensor noise induced by environmental changes, on detection performance. In this paper, we propose MEFormer to address the LiDAR over-reliance problem by harnessing critical information for 3D object detection from every available modality while concurrently safeguarding against corrupted signals during the fusion process. Specifically, we introduce Modality Agnostic Decoding (MOAD) that extracts geometric and semantic features with a shared transformer decoder regardless of input modalities and provides promising improvement with a single modality as well as multi-modality. Additionally, our Proximity-based Modality Ensemble (PME) module adaptively utilizes the strengths of each modality depending on the environment while mitigating the effects of a noisy sensor. Our MEFormer achieves state-of-the-art performance of 73.9% NDS and 71.5% mAP in the nuScenes validation set. Extensive analyses validate that our MEFormer improves robustness against challenging conditions such as sensor malfunctions or environmental changes. The source code is available at https://github.com/hanchaa/MEFormer
Authors: Santosh V. Patapati
Abstract: Major Depressive Disorder (MDD) is a pervasive mental health condition that affects 300 million people worldwide. This work presents a novel, BiLSTM-based tri-modal model-level fusion architecture for the binary classification of depression from clinical interview recordings. The proposed architecture incorporates Mel Frequency Cepstral Coefficients, Facial Action Units, and uses a two-shot learning based GPT-4 model to process text data. This is the first work to incorporate large language models into a multi-modal architecture for this task. It achieves impressive results on the DAIC-WOZ AVEC 2016 Challenge cross-validation split and Leave-One-Subject-Out cross-validation split, surpassing all baseline models and multiple state-of-the-art models. In Leave-One-Subject-Out testing, it achieves an accuracy of 91.01%, an F1-Score of 85.95%, a precision of 80%, and a recall of 92.86%.
Authors: Wenxuan Wang, Quan Sun, Fan Zhang, Yepeng Tang, Jing Liu, Xinlong Wang
Abstract: Contrastive Language-Image Pre-training (CLIP), which excels at abstracting open-world representations across domains and modalities, has become a foundation for a variety of vision and multimodal tasks. However, recent studies reveal that CLIP has severe visual shortcomings, such as which can hardly distinguish orientation, quantity, color, structure, etc. These visual shortcomings also limit the perception capabilities of multimodal large language models (MLLMs) built on CLIP. The main reason could be that the image-text pairs used to train CLIP are inherently biased, due to the lack of the distinctiveness of the text and the diversity of images. In this work, we present a simple post-training approach for CLIP models, which largely overcomes its visual shortcomings via a self-supervised diffusion process. We introduce DIVA, which uses the DIffusion model as a Visual Assistant for CLIP. Specifically, DIVA leverages generative feedback from text-to-image diffusion models to optimize CLIP representations, with only images (without corresponding text). We demonstrate that DIVA improves CLIP's performance on the challenging MMVP-VLM benchmark which assesses fine-grained visual abilities to a large extent (e.g., 3-7%), and enhances the performance of MLLMs and vision models on multimodal understanding and segmentation tasks. Extensive evaluation on 29 image classification and retrieval benchmarks confirms that our framework preserves CLIP's strong zero-shot capabilities. The code is available at https://github.com/baaivision/DIVA.
Authors: Ali Abdollahi, Mahdi Ghaznavi, Mohammad Reza Karimi Nejad, Arash Mari Oriyad, Reza Abbasi, Ali Salesi, Melika Behjati, Mohammad Hossein Rohban, Mahdieh Soleymani Baghshah
Abstract: Vision-language models (VLMs) are intensively used in many downstream tasks, including those requiring assessments of individuals appearing in the images. While VLMs perform well in simple single-person scenarios, in real-world applications, we often face complex situations in which there are persons of different genders doing different activities. We show that in such cases, VLMs are biased towards identifying the individual with the expected gender (according to ingrained gender stereotypes in the model or other forms of sample selection bias) as the performer of the activity. We refer to this bias in associating an activity with the gender of its actual performer in an image or text as the Gender-Activity Binding (GAB) bias and analyze how this bias is internalized in VLMs. To assess this bias, we have introduced the GAB dataset with approximately 5500 AI-generated images that represent a variety of activities, addressing the scarcity of real-world images for some scenarios. To have extensive quality control, the generated images are evaluated for their diversity, quality, and realism. We have tested 12 renowned pre-trained VLMs on this dataset in the context of text-to-image and image-to-text retrieval to measure the effect of this bias on their predictions. Additionally, we have carried out supplementary experiments to quantify the bias in VLMs' text encoders and to evaluate VLMs' capability to recognize activities. Our experiments indicate that VLMs experience an average performance decline of about 13.2% when confronted with gender-activity binding bias.
Authors: Zhixin Yan, Yan Zhong, Hongbin Xu, Wenjun Zhang, Lin Shu, Hongbin Xu, Wenxiong Kang
Abstract: Remote Photoplethysmography (rPPG) is a non-contact technique for extracting physiological signals from facial videos, used in applications like emotion monitoring, medical assistance, and anti-face spoofing. Unlike controlled laboratory settings, real-world environments often contain motion artifacts and noise, affecting the performance of existing rPPG methods. To address this, we propose PhysMamba, a dual-Pathway time-frequency interaction model via State Space Duality. This method allows the network to learn richer, more representative features, enhancing robustness in noisy conditions. To facilitate information exchange and feature complementation between the two pathways, we design an improved algorithm: Cross-Attention State Space Duality (CASSD). We conduct comparative experiments on the PURE, UBFC-rPPG, and MMPD datasets. Experimental results show that PhysMamba achieves state-of-the-art performance, particularly in complex environments, demonstrating its potential in practical remote physiological signal measurement applications.
Authors: Simone Caldarella, Massimiliano Mancini, Elisa Ricci, Rahaf Aljundi
Abstract: Vision-Language Models (VLMs) combine visual and textual understanding, rendering them well-suited for diverse tasks like generating image captions and answering visual questions across various domains. However, these capabilities are built upon training on large amount of uncurated data crawled from the web. The latter may include sensitive information that VLMs could memorize and leak, raising significant privacy concerns. In this paper, we assess whether these vulnerabilities exist, focusing on identity leakage. Our study leads to three key findings: (i) VLMs leak identity information, even when the vision-language alignment and the fine-tuning use anonymized data; (ii) context has little influence on identity leakage; (iii) simple, widely used anonymization techniques, like blurring, are not sufficient to address the problem. These findings underscore the urgent need for robust privacy protection strategies when deploying VLMs. Ethical awareness and responsible development practices are essential to mitigate these risks.
Authors: Zhehan Zhang, Meihua Qian, Li Luo, Ripon Saha, Qianyi Gao, Xinxin Song
Abstract: Assessing artistic creativity has long challenged researchers, with traditional methods proving time-consuming. Recent studies have applied machine learning to evaluate creativity in drawings, but not paintings. Our research addresses this gap by developing a CNN model to automatically assess the creativity of human paintings. Using a dataset of six hundred paintings by professionals and children, our model achieved 90% accuracy and faster evaluation times than human raters. This approach demonstrates the potential of machine learning in advancing artistic creativity assessment, offering a more efficient alternative to traditional methods.
Authors: Royden Wagner, \"Omer Sahin Tas, Marlon Steiner, Fabian Konstantinidis, Hendrik K\"onigshof, Marvin Klemp, Carlos Fernandez, Christoph Stiller
Abstract: Self-driving vehicles rely on multimodal motion forecasts to effectively interact with their environment and plan safe maneuvers. We introduce SceneMotion, an attention-based model for forecasting scene-wide motion modes of multiple traffic agents. Our model transforms local agent-centric embeddings into scene-wide forecasts using a novel latent context module. This module learns a scene-wide latent space from multiple agent-centric embeddings, enabling joint forecasting and interaction modeling. The competitive performance in the Waymo Open Interaction Prediction Challenge demonstrates the effectiveness of our approach. Moreover, we cluster future waypoints in time and space to quantify the interaction between agents. We merge all modes and analyze each mode independently to determine which clusters are resolved through interaction or result in conflict. Our implementation is available at: https://github.com/kit-mrt/future-motion
Authors: Chaolei Tan, Zihang Lin, Junfu Pu, Zhongang Qi, Wei-Yi Pei, Zhi Qu, Yexin Wang, Ying Shan, Wei-Shi Zheng, Jian-Fang Hu
Abstract: Video grounding is a fundamental problem in multimodal content understanding, aiming to localize specific natural language queries in an untrimmed video. However, current video grounding datasets merely focus on simple events and are either limited to shorter videos or brief sentences, which hinders the model from evolving toward stronger multimodal understanding capabilities. To address these limitations, we present a large-scale video grounding dataset named SynopGround, in which more than 2800 hours of videos are sourced from popular TV dramas and are paired with accurately localized human-written synopses. Each paragraph in the synopsis serves as a language query and is manually annotated with precise temporal boundaries in the long video. These paragraph queries are tightly correlated to each other and contain a wealth of abstract expressions summarizing video storylines and specific descriptions portraying event details, which enables the model to learn multimodal perception on more intricate concepts over longer context dependencies. Based on the dataset, we further introduce a more complex setting of video grounding dubbed Multi-Paragraph Video Grounding (MPVG), which takes as input multiple paragraphs and a long video for grounding each paragraph query to its temporal interval. In addition, we propose a novel Local-Global Multimodal Reasoner (LGMR) to explicitly model the local-global structures of long-term multimodal inputs for MPVG. Our method provides an effective baseline solution to the multi-paragraph video grounding problem. Extensive experiments verify the proposed model's effectiveness as well as its superiority in long-term multi-paragraph video grounding over prior state-of-the-arts. Dataset and code are publicly available. Project page: https://synopground.github.io/.
Authors: Ruoyu Wang, Wenqian Wang, Jianjun Gao, Dan Lin, Kim-Hui Yap, Bingbing Li
Abstract: Driver action recognition, aiming to accurately identify drivers' behaviours, is crucial for enhancing driver-vehicle interactions and ensuring driving safety. Unlike general action recognition, drivers' environments are often challenging, being gloomy and dark, and with the development of sensors, various cameras such as IR and depth cameras have emerged for analyzing drivers' behaviors. Therefore, in this paper, we propose a novel multimodal fusion transformer, named MultiFuser, which identifies cross-modal interrelations and interactions among multimodal car cabin videos and adaptively integrates different modalities for improved representations. Specifically, MultiFuser comprises layers of Bi-decomposed Modules to model spatiotemporal features, with a modality synthesizer for multimodal features integration. Each Bi-decomposed Module includes a Modal Expertise ViT block for extracting modality-specific features and a Patch-wise Adaptive Fusion block for efficient cross-modal fusion. Extensive experiments are conducted on Drive&Act dataset and the results demonstrate the efficacy of our proposed approach.
Authors: Junyan Ye, Jun He, Weijia Li, Zhutao Lv, Jinhua Yu, Haote Yang, Conghui He
Abstract: Street-to-satellite image synthesis focuses on generating realistic satellite images from corresponding ground street-view images while maintaining a consistent content layout, similar to looking down from the sky. The significant differences in perspectives create a substantial domain gap between the views, making this cross-view generation task particularly challenging. In this paper, we introduce SkyDiffusion, a novel cross-view generation method for synthesizing satellite images from street-view images, leveraging diffusion models and Bird's Eye View (BEV) paradigm. First, we design a Curved-BEV method to transform street-view images to the satellite view, reformulating the challenging cross-domain image synthesis task into a conditional generation problem. Curved-BEV also includes a "Multi-to-One" mapping strategy for leveraging multiple street-view images within the same satellite coverage area, effectively solving the occlusion issues in dense urban scenes. Next, we design a BEV-controlled diffusion model to generate satellite images consistent with the street-view content, which also incorporates a light manipulation module to make the lighting conditions of the synthesized satellite images more flexible. Experimental results demonstrate that SkyDiffusion outperforms state-of-the-art methods on both suburban (CVUSA & CVACT) and urban (VIGOR-Chicago) cross-view datasets, with an average SSIM increase of 13.96% and a FID reduction of 20.54%, achieving realistic and content-consistent satellite image generation. The code and models of this work will be released at https://opendatalab.github.io/skydiffusion
Authors: Zhiling Yan, Weixiang Sun, Rong Zhou, Zhengqing Yuan, Kai Zhang, Yiwei Li, Tianming Liu, Quanzheng Li, Xiang Li, Lifang He, Lichao Sun
Abstract: Medical image segmentation and video object segmentation are essential for diagnosing and analyzing diseases by identifying and measuring biological structures. Recent advances in natural domain have been driven by foundation models like the Segment Anything Model 2 (SAM-2). To explore the performance of SAM-2 in biomedical applications, we designed three evaluation pipelines for single-frame 2D image segmentation, multi-frame 3D image segmentation and multi-frame video segmentation with varied prompt designs, revealing SAM-2's limitations in medical contexts. Consequently, we developed BioSAM-2, an enhanced foundation model optimized for biomedical data based on SAM-2. Our experiments show that BioSAM-2 not only surpasses the performance of existing state-of-the-art foundation models but also matches or even exceeds specialist models, demonstrating its efficacy and potential in the medical domain.
Authors: Du Chen, Zhengqiang Zhang, Jie Liang, Lei Zhang
Abstract: Generative adversarial networks (GAN) and generative diffusion models (DM) have been widely used in real-world image super-resolution (Real-ISR) to enhance the image perceptual quality. However, these generative models are prone to generating visual artifacts and false image structures, resulting in unnatural Real-ISR results. Based on the fact that natural images exhibit high self-similarities, i.e., a local patch can have many similar patches to it in the whole image, in this work we propose a simple yet effective self-similarity loss (SSL) to improve the performance of generative Real-ISR models, enhancing the hallucination of structural and textural details while reducing the unpleasant visual artifacts. Specifically, we compute a self-similarity graph (SSG) of the ground-truth image, and enforce the SSG of Real-ISR output to be close to it. To reduce the training cost and focus on edge areas, we generate an edge mask from the ground-truth image, and compute the SSG only on the masked pixels. The proposed SSL serves as a general plug-and-play penalty, which could be easily applied to the off-the-shelf Real-ISR models. Our experiments demonstrate that, by coupling with SSL, the performance of many state-of-the-art Real-ISR models, including those GAN and DM based ones, can be largely improved, reproducing more perceptually realistic image details and eliminating many false reconstructions and visual artifacts. Codes and supplementary material can be found at https://github.com/ChrisDud0257/SSL
Authors: Yujia Wu, Yiming Shi, Jiwei Wei, Chengwei Sun, Yuyang Zhou, Yang Yang, Heng Tao Shen
Abstract: Personalized text-to-image generation has gained significant attention for its capability to generate high-fidelity portraits of specific identities conditioned on user-defined prompts. Existing methods typically involve test-time fine-tuning or instead incorporating an additional pre-trained branch. However, these approaches struggle to simultaneously address the demands of efficiency, identity fidelity, and preserving the model's original generative capabilities. In this paper, we propose DiffLoRA, a novel approach that leverages diffusion models as a hypernetwork to predict personalized low-rank adaptation (LoRA) weights based on the reference images. By integrating these LoRA weights into the text-to-image model, DiffLoRA achieves personalization during inference without further training. Additionally, we propose an identity-oriented LoRA weight construction pipeline to facilitate the training of DiffLoRA. By utilizing the dataset produced by this pipeline, our DiffLoRA consistently generates high-performance and accurate LoRA weights. Extensive evaluations demonstrate the effectiveness of our method, achieving both time efficiency and maintaining identity fidelity throughout the personalization process.
Authors: Sean J. Liu, Nupur Kumari, Ariel Shamir, Jun-Yan Zhu
Abstract: Text-to-image models are powerful tools for image creation. However, the generation process is akin to a dice roll and makes it difficult to achieve a single image that captures everything a user wants. In this paper, we propose a framework for creating the desired image by compositing it from various parts of generated images, in essence forming a Generative Photomontage. Given a stack of images generated by ControlNet using the same input condition and different seeds, we let users select desired parts from the generated results using a brush stroke interface. We introduce a novel technique that takes in the user's brush strokes, segments the generated images using a graph-based optimization in diffusion feature space, and then composites the segmented regions via a new feature-space blending method. Our method faithfully preserves the user-selected regions while compositing them harmoniously. We demonstrate that our flexible framework can be used for many applications, including generating new appearance combinations, fixing incorrect shapes and artifacts, and improving prompt alignment. We show compelling results for each application and demonstrate that our method outperforms existing image blending methods and various baselines.
Authors: Hyunjee Lee, Youngsik Yun, Jeongmin Bae, Seoha Kim, Youngjung Uh
Abstract: Understanding the 3D semantics of a scene is a fundamental problem for various scenarios such as embodied agents. While NeRFs and 3DGS excel at novel-view synthesis, previous methods for understanding their semantics have been limited to incomplete 3D understanding: their segmentation results are 2D masks and their supervision is anchored at 2D pixels. This paper revisits the problem set to pursue a better 3D understanding of a scene modeled by NeRFs and 3DGS as follows. 1) We directly supervise the 3D points to train the language embedding field. It achieves state-of-the-art accuracy without relying on multi-scale language embeddings. 2) We transfer the pre-trained language field to 3DGS, achieving the first real-time rendering speed without sacrificing training time or accuracy. 3) We introduce a 3D querying and evaluation protocol for assessing the reconstructed geometry and semantics together. Code, checkpoints, and annotations will be available online. Project page: https://hyunji12.github.io/Open3DRF
Authors: Yibin Wang, Weizhong Zhang, Cheng Jin
Abstract: Current state-of-the-art methods for human image customized synthesis typically require tedious training on large-scale datasets. In such cases, they are prone to overfitting and struggle to personalize individuals of unseen styles. Moreover, these methods extensively focus on single-concept human image synthesis and lack the flexibility needed for customizing individuals with multiple given concepts, thereby impeding their broader practical application. To this end, we propose MagicFace, a novel training-free method for universal-style human image personalized synthesis, enabling multi-concept customization by accurately integrating reference concept features into their latent generated region at the pixel level. Specifically, MagicFace introduces a coarse-to-fine generation pipeline, involving two sequential stages: semantic layout construction and concept feature injection. This is achieved by our Reference-aware Self-Attention (RSA) and Region-grouped Blend Attention (RBA) mechanisms. In the first stage, RSA enables the latent image to query features from all reference concepts simultaneously, extracting the overall semantic understanding to facilitate the initial semantic layout establishment. In the second stage, we employ an attention-based semantic segmentation method to pinpoint the latent generated regions of all concepts at each step. Following this, RBA divides the pixels of the latent image into semantic groups, with each group querying fine-grained features from the corresponding reference concept, which ensures precise attribute alignment and feature injection. Throughout the generation process, a weighted mask strategy is employed to ensure the model focuses more on the reference concepts. Extensive experiments demonstrate the superiority of MagicFace in both human-centric subject-to-image synthesis and multi-concept human image customization.
Authors: Wujie Sun, Defang Chen, Siwei Lyu, Genlang Chen, Chun Chen, Can Wang
Abstract: Recent research on knowledge distillation has increasingly focused on logit distillation because of its simplicity, effectiveness, and versatility in model compression. In this paper, we introduce Refined Logit Distillation (RLD) to address the limitations of current logit distillation methods. Our approach is motivated by the observation that even high-performing teacher models can make incorrect predictions, creating a conflict between the standard distillation loss and the cross-entropy loss. This conflict can undermine the consistency of the student model's learning objectives. Previous attempts to use labels to empirically correct teacher predictions may undermine the class correlation. In contrast, our RLD employs labeling information to dynamically refine teacher logits. In this way, our method can effectively eliminate misleading information from the teacher while preserving crucial class correlations, thus enhancing the value and efficiency of distilled knowledge. Experimental results on CIFAR-100 and ImageNet demonstrate its superiority over existing methods. The code is provided at \text{https://github.com/zju-SWJ/RLD}.
Authors: Guofeng Feng, Siyan Chen, Rong Fu, Zimu Liao, Yi Wang, Tao Liu, Zhilin Pei, Hengjie Li, Xingcheng Zhang, Bo Dai
Abstract: This work introduces FlashGS, an open-source CUDA Python library, designed to facilitate the efficient differentiable rasterization of 3D Gaussian Splatting through algorithmic and kernel-level optimizations. FlashGS is developed based on the observations from a comprehensive analysis of the rendering process to enhance computational efficiency and bring the technique to wide adoption. The paper includes a suite of optimization strategies, encompassing redundancy elimination, efficient pipelining, refined control and scheduling mechanisms, and memory access optimizations, all of which are meticulously integrated to amplify the performance of the rasterization process. An extensive evaluation of FlashGS' performance has been conducted across a diverse spectrum of synthetic and real-world large-scale scenes, encompassing a variety of image resolutions. The empirical findings demonstrate that FlashGS consistently achieves an average 4x acceleration over mobile consumer GPUs, coupled with reduced memory consumption. These results underscore the superior performance and resource optimization capabilities of FlashGS, positioning it as a formidable tool in the domain of 3D rendering.
Authors: Shuai Yuan, Hanlin Qin, Renke Kou, Xiang Yan, Zechuan Li, Chenxu Peng, Abd-Krim Seghouane
Abstract: In this work, we make the first attempt to construct a learning-based single-point annotation paradigm for infrared small target label generation (IRSTLG). Our intuition is that label generation requires just one more point prompt than target detection: IRSTLG can be regarded as an infrared small target detection (IRSTD) task with the target location hint. Based on this insight, we introduce an energy double guided single-point prompt (EDGSP) framework, which adeptly transforms the target detection network into a refined label generation method. Specifically, the proposed EDGSP includes: 1) target energy initialization (TEI) to create a foundational outline for sufficient shape evolution of pseudo label, 2) double prompt embedding (DPE) for rapid localization of interested regions and reinforcement of individual differences to avoid label adhesion, and 3) bounding box-based matching (BBM) to eliminate false alarms. Experimental results show that pseudo labels generated by three baselines equipped with EDGSP achieve 100% object-level probability of detection (Pd) and 0% false-alarm rate (Fa) on SIRST, NUDT-SIRST, and IRSTD-1k datasets, with a pixel-level intersection over union (IoU) improvement of 13.28% over state-of-the-art (SOTA) label generation methods. In the practical application of downstream IRSTD, EDGSP realizes, for the first time, a single-point generated pseudo mask beyond the full label. Even with coarse single-point annotations, it still achieves 99.5% performance of full labeling.
Authors: Chen Long-fei, Yuichi Nakamura, Kazuaki Kondo
Abstract: With respect to machine operation tasks, the experiences from different skill level operators, especially novices, can provide worthy understanding about the manner in which they perceive the operational environment and formulate knowledge to deal with various operation situations. In this study, we describe the operator's behaviors by utilizing the relations among their head, hand, and operation location (hotspot) during the operation. A total of 40 experiences associated with a sewing machine operation task performed by amateur operators was recorded via a head-mounted RGB-D camera. We examined important features of operational behaviors in different skill level operators and confirmed their correlation to the difficulties of the operation steps. The result shows that the pure-gazing behavior is significantly reduced when the operator's skill improved. Moreover, the hand-approaching duration and the frequency of attention movement before operation are strongly correlated to the operational difficulty in such machine operating environments.
Authors: Max Vladymyrov, Andrey Zhmoginov, Mark Sandler
Abstract: We focus on the problem of learning without forgetting from multiple tasks arriving sequentially, where each task is defined using a few-shot episode of novel or already seen classes. We approach this problem using the recently published HyperTransformer (HT), a Transformer-based hypernetwork that generates specialized task-specific CNN weights directly from the support set. In order to learn from a continual sequence of tasks, we propose to recursively re-use the generated weights as input to the HT for the next task. This way, the generated CNN weights themselves act as a representation of previously learned tasks, and the HT is trained to update these weights so that the new task can be learned without forgetting past tasks. This approach is different from most continual learning algorithms that typically rely on using replay buffers, weight regularization or task-dependent architectural changes. We demonstrate that our proposed Continual HyperTransformer method equipped with a prototypical loss is capable of learning and retaining knowledge about past tasks for a variety of scenarios, including learning from mini-batches, and task-incremental and class-incremental learning scenarios.
Authors: Ling Huang, Su Ruan, Pierre Decazes, Thierry Denoeux
Abstract: Single-modality medical images generally do not contain enough information to reach an accurate and reliable diagnosis. For this reason, physicians generally diagnose diseases based on multimodal medical images such as, e.g., PET/CT. The effective fusion of multimodal information is essential to reach a reliable decision and explain how the decision is made as well. In this paper, we propose a fusion framework for multimodal medical image segmentation based on deep learning and the Dempster-Shafer theory of evidence. In this framework, the reliability of each single modality image when segmenting different objects is taken into account by a contextual discounting operation. The discounted pieces of evidence from each modality are then combined by Dempster's rule to reach a final decision. Experimental results with a PET-CT dataset with lymphomas and a multi-MRI dataset with brain tumors show that our method outperforms the state-of-the-art methods in accuracy and reliability.
Authors: Surbhi Mittal, Kartik Thakral, Richa Singh, Mayank Vatsa, Tamar Glaser, Cristian Canton Ferrer, Tal Hassner
Abstract: Artificial Intelligence (AI) has made its way into various scientific fields, providing astonishing improvements over existing algorithms for a wide variety of tasks. In recent years, there have been severe concerns over the trustworthiness of AI technologies. The scientific community has focused on the development of trustworthy AI algorithms. However, machine and deep learning algorithms, popular in the AI community today, depend heavily on the data used during their development. These learning algorithms identify patterns in the data, learning the behavioral objective. Any flaws in the data have the potential to translate directly into algorithms. In this study, we discuss the importance of Responsible Machine Learning Datasets and propose a framework to evaluate the datasets through a responsible rubric. While existing work focuses on the post-hoc evaluation of algorithms for their trustworthiness, we provide a framework that considers the data component separately to understand its role in the algorithm. We discuss responsible datasets through the lens of fairness, privacy, and regulatory compliance and provide recommendations for constructing future datasets. After surveying over 100 datasets, we use 60 datasets for analysis and demonstrate that none of these datasets is immune to issues of fairness, privacy preservation, and regulatory compliance. We provide modifications to the ``datasheets for datasets" with important additions for improved dataset documentation. With governments around the world regularizing data protection laws, the method for the creation of datasets in the scientific community requires revision. We believe this study is timely and relevant in today's era of AI.
Authors: Naoki Wake, Atsushi Kanehira, Kazuhiro Sasabuchi, Jun Takamatsu, Katsushi Ikeuchi
Abstract: We introduce a pipeline that enhances a general-purpose Vision Language Model, GPT-4V(ision), to facilitate one-shot visual teaching for robotic manipulation. This system analyzes videos of humans performing tasks and outputs executable robot programs that incorporate insights into affordances. The process begins with GPT-4V analyzing the videos to obtain textual explanations of environmental and action details. A GPT-4-based task planner then encodes these details into a symbolic task plan. Subsequently, vision systems spatially and temporally ground the task plan in the videos. Object are identified using an open-vocabulary object detector, and hand-object interactions are analyzed to pinpoint moments of grasping and releasing. This spatiotemporal grounding allows for the gathering of affordance information (e.g., grasp types, waypoints, and body postures) critical for robot execution. Experiments across various scenarios demonstrate the method's efficacy in achieving real robots' operations from human demonstrations in a one-shot manner. Meanwhile, quantitative tests have revealed instances of hallucination in GPT-4V, highlighting the importance of incorporating human supervision within the pipeline. The prompts of GPT-4V/GPT-4 are available at this project page: https://microsoft.github.io/GPT4Vision-Robot-Manipulation-Prompts/
URLs: https://microsoft.github.io/GPT4Vision-Robot-Manipulation-Prompts/
Authors: Shashank Kotyan, Tatsuya Ueda, Danilo Vasconcellos Vargas
Abstract: Most examinations of neural networks' learned latent spaces typically employ dimensionality reduction techniques such as t-SNE or UMAP. These methods distort the local neighborhood in the visualization, making it hard to distinguish the structure of a subset of samples in the latent space. In response to this challenge, we introduce the {k*~distribution} and its corresponding visualization technique This method uses local neighborhood analysis to guarantee the preservation of the structure of sample distributions for individual classes within the subset of the learned latent space. This facilitates easy comparison of different k*~distributions, enabling analysis of how various classes are processed by the same neural network. Our study reveals three distinct distributions of samples within the learned latent space subset: a) Fractured, b) Overlapped, and c) Clustered, providing a more profound understanding of existing contemporary visualizations. Experiments show that the distribution of samples within the network's learned latent space significantly varies depending on the class. Furthermore, we illustrate that our analysis can be applied to explore the latent space of diverse neural network architectures, various layers within neural networks, transformations applied to input samples, and the distribution of training and testing data for neural networks. Thus, the k* distribution should aid in visualizing the structure inside neural networks and further foster their understanding. Project Website is available online at https://shashankkotyan.github.io/k-Distribution/.
Authors: Kuangpu Guo, Yuhe Ding, Jian Liang, Ran He, Zilei Wang, Tieniu Tan
Abstract: Label skews, characterized by disparities in local label distribution across clients, pose a significant challenge in federated learning. As minority classes suffer from worse accuracy due to overfitting on local imbalanced data, prior methods often incorporate class-balanced learning techniques during local training. Although these methods improve the mean accuracy across all classes, we observe that vacant classes-referring to categories absent from a client's data distribution-remain poorly recognized. Besides, there is still a gap in the accuracy of local models on minority classes compared to the global model. This paper introduces FedVLS, a novel approach to label-skewed federated learning that integrates both vacant-class distillation and logit suppression simultaneously. Specifically, vacant-class distillation leverages knowledge distillation during local training on each client to retain essential information related to vacant classes from the global model. Moreover, logit suppression directly penalizes network logits for non-label classes, effectively addressing misclassifications in minority classes that may be biased toward majority classes. Extensive experiments validate the efficacy of FedVLS, demonstrating superior performance compared to previous state-of-the-art (SOTA) methods across diverse datasets with varying degrees of label skews. Code is available in the supplementary material.
Authors: Minheng Chen, Tonglong Li, Zhirun Zhang, Youyong Kong
Abstract: A robust and efficient optimization-based 2D/3D registration framework is crucial for the navigation system of orthopedic surgical robots. It can provide precise position information of surgical instruments and implants during surgery. While artificial intelligence technology has advanced rapidly in recent years, traditional optimization-based registration methods remain indispensable in the field of 2D/3D registration.he exceptional precision of this method enables it to be considered as a post-processing step of the learning-based methods, thereby offering a reliable assurance for registration. In this paper, we present a coarse-to-fine registration framework based on the CMA-ES algorithm. We conducted intensive testing of our method using data from different parts of the spine. The results shows the effectiveness of the proposed framework on real orthopedic spine surgery clinical data. This work can be viewed as an additional extension that complements the optimization-based methods employed in our previous studies.
Authors: Ben Pinhasov, Raz Lapid, Rony Ohayon, Moshe Sipper, Yehudit Aperstein
Abstract: We introduce a novel methodology for identifying adversarial attacks on deepfake detectors using eXplainable Artificial Intelligence (XAI). In an era characterized by digital advancement, deepfakes have emerged as a potent tool, creating a demand for efficient detection systems. However, these systems are frequently targeted by adversarial attacks that inhibit their performance. We address this gap, developing a defensible deepfake detector by leveraging the power of XAI. The proposed methodology uses XAI to generate interpretability maps for a given method, providing explicit visualizations of decision-making factors within the AI models. We subsequently employ a pretrained feature extractor that processes both the input image and its corresponding XAI image. The feature embeddings extracted from this process are then used for training a simple yet effective classifier. Our approach contributes not only to the detection of deepfakes but also enhances the understanding of possible adversarial attacks, pinpointing potential vulnerabilities. Furthermore, this approach does not change the performance of the deepfake detector. The paper demonstrates promising results suggesting a potential pathway for future deepfake detection mechanisms. We believe this study will serve as a valuable contribution to the community, sparking much-needed discourse on safeguarding deepfake detectors.
Authors: Patryk Rygiel, Dieuwertje Alblas, Christoph Brune, Kak Khee Yeung, Jelmer M. Wolterink
Abstract: Personalized 3D vascular models can aid in a range of diagnostic, prognostic, and treatment-planning tasks relevant to cardiovascular disease management. Deep learning provides a means to obtain such models automatically from image data. Ideally, a user should have control over the included region in the vascular model. Additionally, the model should be watertight and highly accurate. To this end, we propose a combination of a global controller leveraging voxel mask segmentations to provide boundary conditions for vessels of interest to a local, iterative vessel segmentation model. We introduce the preservation of scale- and rotational symmetries in the local segmentation model, leading to generalisation to vessels of unseen sizes and orientations. Combined with the global controller, this enables flexible 3D vascular model building, without additional retraining. We demonstrate the potential of our method on a dataset containing abdominal aortic aneurysms (AAAs). Our method performs on par with a state-of-the-art segmentation model in the segmentation of AAAs, iliac arteries, and renal arteries, while providing a watertight, smooth surface representation. Moreover, we demonstrate that by adapting the global controller, we can easily extend vessel sections in the 3D model.
Authors: Jiayu Huo, Ruiqiang Xiao, Haotian Zheng, Yang Liu, Sebastien Ourselin, Rachel Sparks
Abstract: Recently, automated medical image segmentation methods based on deep learning have achieved great success. However, they heavily rely on large annotated datasets, which are costly and time-consuming to acquire. Few-shot learning aims to overcome the need for annotated data by using a small labeled dataset, known as a support set, to guide predicting labels for new, unlabeled images, known as the query set. Inspired by this paradigm, we introduce MatchSeg, a novel framework that enhances medical image segmentation through strategic reference image matching. We leverage contrastive language-image pre-training (CLIP) to select highly relevant samples when defining the support set. Additionally, we design a joint attention module to strengthen the interaction between support and query features, facilitating a more effective knowledge transfer between support and query sets. We validated our method across four public datasets. Experimental results demonstrate superior segmentation performance and powerful domain generalization ability of MatchSeg against existing methods for domain-specific and cross-domain segmentation tasks. Our code is made available at https://github.com/keeplearning-again/MatchSeg
Authors: Yingtao Shen, Minqing Sun, Jianzhe Lin, Jie Zhao, An Zou
Abstract: Model compression has gained significant popularity as a means to alleviate the computational and memory demands of machine learning models. Each compression technique leverages unique features to reduce the size of neural networks. Although intuitively combining different techniques may enhance compression effectiveness, we find that the order in which they are combined significantly influences performance. To identify the optimal sequence for compressing neural networks, we propose the Order of Compression, a systematic and optimal sequence to apply multiple compression techniques in the most effective order. We start by building the foundations of the orders between any two compression approaches and then demonstrate inserting additional compression between any two compressions will not break the order of the two compression approaches. Based on the foundations, an optimal order is obtained with topological sorting. Validated on image-based regression and classification networks across different datasets, our proposed Order of Compression significantly reduces computational costs by up to 859 times on ResNet34, with negligible accuracy loss (-0.09% for CIFAR10) compared to the baseline model. We believe our simple yet effective exploration of the order of compression will shed light on the practice of model compression.
Authors: Zhenhuan Zhou, Along He, Yanlin Wu, Rui Yao, Xueshuo Xie, Tao Li
Abstract: In medical images, various types of lesions often manifest significant differences in their shape and texture. Accurate medical image segmentation demands deep learning models with robust capabilities in multi-scale and boundary feature learning. However, previous networks still have limitations in addressing the above issues. Firstly, previous networks simultaneously fuse multi-level features or employ deep supervision to enhance multi-scale learning. However, this may lead to feature redundancy and excessive computational overhead, which is not conducive to network training and clinical deployment. Secondly, the majority of medical image segmentation networks exclusively learn features in the spatial domain, disregarding the abundant global information in the frequency domain. This results in a bias towards low-frequency components, neglecting crucial high-frequency information. To address these problems, we introduce SF-UNet, a spatial-frequency dual-domain attention network. It comprises two main components: the Multi-scale Progressive Channel Attention (MPCA) block, which progressively extract multi-scale features across adjacent encoder layers, and the lightweight Frequency-Spatial Attention (FSA) block, with only 0.05M parameters, enabling concurrent learning of texture and boundary features from both spatial and frequency domains. We validate the effectiveness of the proposed SF-UNet on three public datasets. Experimental results show that compared to previous state-of-the-art (SOTA) medical image segmentation networks, SF-UNet achieves the best performance, and achieves up to 9.4\% and 10.78\% improvement in DSC and IOU. Codes will be released at https://github.com/nkicsl/SF-UNet.
Authors: Chao Fan, Hongyuan Yu, Yan Huang, Liang Wang, Zhenghan Yang, Xibin Jia
Abstract: Despite the progress made in Mamba-based medical image segmentation models, existing methods utilizing unidirectional or multi-directional feature scanning mechanisms struggle to effectively capture dependencies between neighboring positions, limiting the discriminant representation learning of local features. These local features are crucial for medical image segmentation as they provide critical structural information about lesions and organs. To address this limitation, we propose SliceMamba, a simple and effective locally sensitive Mamba-based medical image segmentation model. SliceMamba includes an efficient Bidirectional Slice Scan module (BSS), which performs bidirectional feature slicing and employs varied scanning mechanisms for sliced features with distinct shapes. This design ensures that spatially adjacent features remain close in the scanning sequence, thereby improving segmentation performance. Additionally, to fit the varying sizes and shapes of lesions and organs, we further introduce an Adaptive Slice Search method to automatically determine the optimal feature slice method based on the characteristics of the target data. Extensive experiments on two skin lesion datasets (ISIC2017 and ISIC2018), two polyp segmentation (Kvasir and ClinicDB) datasets, and one multi-organ segmentation dataset (Synapse) validate the effectiveness of our method.
Authors: Qing Xu, Jiaxuan Li, Xiangjian He, Ziyu Liu, Zhen Chen, Wenting Duan, Chenxin Li, Maggie M. He, Fiseha B. Tesema, Wooi P. Cheah, Yi Wang, Rong Qu, Jonathan M. Garibaldi
Abstract: The universality of deep neural networks across different modalities and their generalization capabilities to unseen domains play an essential role in medical image segmentation. The recent Segment Anything Model (SAM) has demonstrated its potential in both settings. However, the huge computational costs, demand for manual annotations as prompts and conflict-prone decoding process of SAM degrade its generalizability and applicability in clinical scenarios. To address these issues, we propose an efficient self-prompting SAM for universal domain-generalized medical image segmentation, named ESP-MedSAM. Specifically, we first devise the Multi-Modal Decoupled Knowledge Distillation (MMDKD) strategy to construct a lightweight semi-parameter sharing image encoder that produces discriminative visual features for diverse modalities. Further, we introduce the Self-Patch Prompt Generator (SPPG) to automatically generate high-quality dense prompt embeddings for guiding segmentation decoding. Finally, we design the Query-Decoupled Modality Decoder (QDMD) that leverages a one-to-one strategy to provide an independent decoding channel for every modality. Extensive experiments indicate that ESP-MedSAM outperforms state-of-the-arts in diverse medical imaging segmentation tasks, displaying superior modality universality and generalization capabilities. Especially, ESP-MedSAM uses only 4.5\% parameters compared to SAM-H. The source code is available at https://github.com/xq141839/ESP-MedSAM.
Authors: R\'ois\'in Luo, James McDermott, Colm O'Riordan
Abstract: Perturbation robustness evaluates the vulnerabilities of models, arising from a variety of perturbations, such as data corruptions and adversarial attacks. Understanding the mechanisms of perturbation robustness is critical for global interpretability. We present a model-agnostic, global mechanistic interpretability method to interpret the perturbation robustness of image models. This research is motivated by two key aspects. First, previous global interpretability works, in tandem with robustness benchmarks, e.g. mean corruption error (mCE), are not designed to directly interpret the mechanisms of perturbation robustness within image models. Second, we notice that the spectral signal-to-noise ratios (SNR) of perturbed natural images exponentially decay over the frequency. This power-law-like decay implies that: Low-frequency signals are generally more robust than high-frequency signals -- yet high classification accuracy can not be achieved by low-frequency signals alone. By applying Shapley value theory, our method axiomatically quantifies the predictive powers of robust features and non-robust features within an information theory framework. Our method, dubbed as \textbf{I-ASIDE} (\textbf{I}mage \textbf{A}xiomatic \textbf{S}pectral \textbf{I}mportance \textbf{D}ecomposition \textbf{E}xplanation), provides a unique insight into model robustness mechanisms. We conduct extensive experiments over a variety of vision models pre-trained on ImageNet to show that \textbf{I-ASIDE} can not only \textbf{measure} the perturbation robustness but also \textbf{provide interpretations} of its mechanisms.
Authors: Qilei Li, Ahmed M. Abdelmoniem
Abstract: Federated Learning (FL) is a distributed machine learning diagram that enables multiple clients to collaboratively train a global model without sharing their private local data. However, FL systems are vulnerable to attacks that are happening in malicious clients through data poisoning and model poisoning, which can deteriorate the performance of aggregated global model. Existing defense methods typically focus on mitigating specific types of poisoning and are often ineffective against unseen types of attack. These methods also assume an attack happened moderately while is not always holds true in real. Consequently, these methods can significantly fail in terms of accuracy and robustness when detecting and addressing updates from attacked malicious clients. To overcome these challenges, in this work, we propose a simple yet effective framework to detect malicious clients, namely Confidence-Aware Defense (CAD), that utilizes the confidence scores of local models as criteria to evaluate the reliability of local updates. Our key insight is that malicious attacks, regardless of attack type, will cause the model to deviate from its previous state, thus leading to increased uncertainty when making predictions. Therefore, CAD is comprehensively effective for both model poisoning and data poisoning attacks by accurately identifying and mitigating potential malicious updates, even under varying degrees of attacks and data heterogeneity. Experimental results demonstrate that our method significantly enhances the robustness of FL systems against various types of attacks across various scenarios by achieving higher model accuracy and stability.
Authors: Viet Anh Nguyen, Minh Lenhat, Khoa Nguyen, Duong Duc Hieu, Dao Huu Hung, Truong Son Hy
Abstract: The versatility of self-attention mechanism earned transformers great success in almost all data modalities, with limitations on the quadratic complexity and difficulty of training. To apply transformers across different data modalities, practitioners have to make specific clever data-modality-dependent constructions. In this paper, we propose Sampling Foundational Transformer (SFT) that can work on multiple data modalities (e.g., point cloud, graph, and sequence) and constraints (e.g., rotational-invariant). The existence of such model is important as contemporary foundational modeling requires operability on multiple data sources. For efficiency on large number of tokens, our model relies on our context aware sampling-without-replacement mechanism for both linear asymptotic computational complexity and real inference time gain. For efficiency, we rely on our newly discovered pseudoconvex formulation of transformer layer to increase model's convergence rate. As a model working on multiple data modalities, SFT has achieved competitive results on many benchmarks, while being faster in inference, compared to other very specialized models.
Authors: Royina Karegoudra Jayanth, Yinshuang Xu, Ziyun Wang, Evangelos Chatzipantazis, Daniel Gehrig, Kostas Daniilidis
Abstract: Neural networks are seeing rapid adoption in purely inertial odometry, where accelerometer and gyroscope measurements from commodity inertial measurement units (IMU) are used to regress displacements and associated uncertainties. They can learn informative displacement priors, which can be directly fused with the raw data with off-the-shelf non-linear filters. Nevertheless, these networks do not consider the physical roto-reflective symmetries inherent in IMU data, leading to the need to memorize the same priors for every possible motion direction, which hinders generalization. In this work, we characterize these symmetries and show that the IMU data and the resulting displacement and covariance transform equivariantly, when rotated around the gravity vector and reflected with respect to arbitrary planes parallel to gravity. We design a neural network that respects these symmetries by design through equivariant processing in three steps: First, it estimates an equivariant gravity-aligned frame from equivariant vectors and invariant scalars derived from IMU data, leveraging expressive linear and non-linear layers tailored to commute with the underlying symmetry transformation. We then map the IMU data into this frame, thereby achieving an invariant canonicalization that can be directly used with off-the-shelf inertial odometry networks. Finally, we map these network outputs back into the original frame, thereby obtaining equivariant covariances and displacements. We demonstrate the generality of our framework by applying it to the filter-based approach based on TLIO, and the end-to-end RONIN architecture, and show better performance on the TLIO, Aria, RIDI and OxIOD datasets than existing methods.
Authors: Wenhao Li, Jie Zhou, Chuan Luo, Chao Tang, Kun Zhang, Shixiong Zhao
Abstract: In the realm of modern mobile E-commerce, providing users with nearby commercial service recommendations through location-based online services has become increasingly vital. While machine learning approaches have shown promise in multi-scene recommendation, existing methodologies often struggle to address cold-start problems in unprecedented scenes: the increasing diversity of commercial choices, along with the short online lifespan of scenes, give rise to the complexity of effective recommendations in online and dynamic scenes. In this work, we propose Scene-wise Adaptive Network (SwAN), a novel approach that emphasizes high-performance cold-start online recommendations for new scenes. Our approach introduces several crucial capabilities, including scene similarity learning, user-specific scene transition cognition, scene-specific information construction for the new scene, and enhancing the diverged logical information between scenes. We demonstrate SwAN's potential to optimize dynamic multi-scene recommendation problems by effectively online handling cold-start recommendations for any newly arrived scenes. More encouragingly, SwAN has been successfully deployed in Meituan's online catering recommendation service, which serves millions of customers per day, and SwAN has achieved a 5.64% CTR index improvement relative to the baselines and a 5.19% increase in daily order volume proportion.