Authors: Hritik Bansal, Zongyu Lin, Tianyi Xie, Zeshun Zong, Michal Yarom, Yonatan Bitton, Chenfanfu Jiang, Yizhou Sun, Kai-Wei Chang, Aditya Grover
Abstract: Recent advances in internet-scale video data pretraining have led to the development of text-to-video generative models that can create high-quality videos across a broad range of visual concepts and styles. Due to their ability to synthesize realistic motions and render complex objects, these generative models have the potential to become general-purpose simulators of the physical world. However, it is unclear how far we are from this goal with the existing text-to-video generative models. To this end, we present VideoPhy, a benchmark designed to assess whether the generated videos follow physical commonsense for real-world activities (e.g. marbles will roll down when placed on a slanted surface). Specifically, we curate a list of 688 captions that involve interactions between various material types in the physical world (e.g., solid-solid, solid-fluid, fluid-fluid). We then generate videos conditioned on these captions from diverse state-of-the-art text-to-video generative models, including open models (e.g., VideoCrafter2) and closed models (e.g., Lumiere from Google, Pika). Further, our human evaluation reveals that the existing models severely lack the ability to generate videos adhering to the given text prompts, while also lack physical commonsense. Specifically, the best performing model, Pika, generates videos that adhere to the caption and physical laws for only 19.7% of the instances. VideoPhy thus highlights that the video generative models are far from accurately simulating the physical world. Finally, we also supplement the dataset with an auto-evaluator, VideoCon-Physics, to assess semantic adherence and physical commonsense at scale.
Authors: Utsab Saha, Sawradip Saha, Shaikh Anowarul Fattah, Mohammad Saquib
Abstract: The identification and restoration of ancient watermarks have long been a major topic in codicology and history. Classifying historical documents based on watermarks can be difficult due to the diversity of watermarks, crowded and noisy samples, multiple modes of representation, and minor distinctions between classes and intra-class changes. This paper proposes a U-net-based conditional generative adversarial network (GAN) to translate noisy raw historical watermarked images into clean, handwriting-free images with just watermarks. Considering its ability to perform image translation from degraded (noisy) pixels to clean pixels, the proposed network is termed as Npix2Cpix. Instead of employing directly degraded watermarked images, the proposed network uses image-to-image translation using adversarial learning to create clutter and handwriting-free images for restoring and categorizing the watermarks for the first time. In order to learn the mapping from input noisy image to output clean image, the generator and discriminator of the proposed U-net-based GAN are trained using two separate loss functions, each of which is based on the distance between images. After using the proposed GAN to pre-process noisy watermarked images, Siamese-based one-shot learning is used to classify watermarks. According to experimental results on a large-scale historical watermark dataset, extracting watermarks from tainted images can result in high one-shot classification accuracy. The qualitative and quantitative evaluation of the retrieved watermarks illustrates the effectiveness of the proposed approach.
Authors: Ulan Alsiyeu, Zhasdauren Duisebekov
Abstract: This paper tackles critical challenges in traffic sign recognition (TSR), which is essential for road safety -- specifically, class imbalance and instance scarcity in datasets. We introduce tailored data augmentation techniques, including synthetic image generation, geometric transformations, and a novel obstacle-based augmentation method to enhance dataset quality for improved model robustness and accuracy. Our methodology incorporates diverse augmentation processes to accurately simulate real-world conditions, thereby expanding the training data's variety and representativeness. Our findings demonstrate substantial improvements in TSR models performance, offering significant implications for traffic sign recognition systems. This research not only addresses dataset limitations in TSR but also proposes a model for similar challenges across different regions and applications, marking a step forward in the field of computer vision and traffic sign recognition systems.
Authors: E. Zhixuan Zeng, Yuhao Chen, Alexander Wong
Abstract: Image generation techniques, particularly latent diffusion models, have exploded in popularity in recent years. Many techniques have been developed to manipulate and clarify the semantic concepts these large-scale models learn, offering crucial insights into biases and concept relationships. However, these techniques are often only validated in conventional realms of human or animal faces and artistic style transitions. The food domain offers unique challenges through complex compositions and regional biases, which can shed light on the limitations and opportunities within existing methods. Through the lens of food imagery, we analyze both qualitative and quantitative patterns within a concept traversal technique. We reveal measurable insights into the model's ability to capture and represent the nuances of culinary diversity, while also identifying areas where the model's biases and limitations emerge.
Authors: Harshvardhan Mestha, Tejas Agarwal, Karan Bania, Shreyas V, Yash Bhisikar
Abstract: Large vision-language models (VLMs) are shown to learn rich joint image-text representations enabling high performances in relevant downstream tasks. However, they fail to showcase their quantitative understanding of objects, and they lack good counting-aware representation. This paper conducts a reproducibility study of 'Teaching CLIP to Count to Ten' (Paiss et al., 2023), which presents a method to finetune a CLIP model (Radford et al., 2021) to improve zero-shot counting accuracy in an image while maintaining the performance for zero-shot classification by introducing a counting-contrastive loss term. We improve the model's performance on a smaller subset of their training data with lower computational resources. We verify these claims by reproducing their study with our own code. The implementation can be found at https://github.com/SforAiDl/CountCLIP.
Authors: Masum Hasan, Cengiz Ozel, Nina Long, Alexander Martin, Samuel Potter, Tariq Adnan, Sangwu Lee, Amir Zadeh, Ehsan Hoque
Abstract: We propose a new large synthetic hand pose estimation dataset, Hi5, and a novel inexpensive method for collecting high-quality synthetic data that requires no human annotation or validation. Leveraging recent advancements in computer graphics, high-fidelity 3D hand models with diverse genders and skin colors, and dynamic environments and camera movements, our data synthesis pipeline allows precise control over data diversity and representation, ensuring robust and fair model training. We generate a dataset with 583,000 images with accurate pose annotation using a single consumer PC that closely represents real-world variability. Pose estimation models trained with Hi5 perform competitively on real-hand benchmarks while surpassing models trained with real data when tested on occlusions and perturbations. Our experiments show promising results for synthetic data as a viable solution for data representation problems in real datasets. Overall, this paper provides a promising new approach to synthetic data creation and annotation that can reduce costs and increase the diversity and quality of data for hand pose estimation.
Authors: Yan Zhang, Sergey Prokudin, Marko Mihajlovic, Qianli Ma, Siyu Tang
Abstract: Understanding the dynamics of generic 3D scenes is fundamentally challenging in computer vision, essential in enhancing applications related to scene reconstruction, motion tracking, and avatar creation. In this work, we address the task as the problem of inferring dense, long-range motion of 3D points. By observing a set of point trajectories, we aim to learn an implicit motion field parameterized by a neural network to predict the movement of novel points within the same domain, without relying on any data-driven or scene-specific priors. To achieve this, our approach builds upon the recently introduced dynamic point field model that learns smooth deformation fields between the canonical frame and individual observation frames. However, temporal consistency between consecutive frames is neglected, and the number of required parameters increases linearly with the sequence length due to per-frame modeling. To address these shortcomings, we exploit the intrinsic regularization provided by SIREN, and modify the input layer to produce a spatiotemporally smooth motion field. Additionally, we analyze the motion field Jacobian matrix, and discover that the motion degrees of freedom (DOFs) in an infinitesimal area around a point and the network hidden variables have different behaviors to affect the model's representational power. This enables us to improve the model representation capability while retaining the model compactness. Furthermore, to reduce the risk of overfitting, we introduce a regularization term based on the assumption of piece-wise motion smoothness. Our experiments assess the model's performance in predicting unseen point trajectories and its application in temporal mesh alignment with guidance. The results demonstrate its superiority and effectiveness. The code and data for the project are publicly available: \url{https://yz-cnsdqz.github.io/eigenmotion/DOMA/}
Authors: Behzad Vahedi, Benjamin Lucas, Farnoush Banaei-Kashani, Andrew P. Barrett, Walter N. Meier, Siri Jodha Khalsa, Morteza Karimzadeh
Abstract: Sea ice, crucial to the Arctic and Earth's climate, requires consistent monitoring and high-resolution mapping. Manual sea ice mapping, however, is time-consuming and subjective, prompting the need for automated deep learning-based classification approaches. However, training these algorithms is challenging because expert-generated ice charts, commonly used as training data, do not map single ice types but instead map polygons with multiple ice types. Moreover, the distribution of various ice types in these charts is frequently imbalanced, resulting in a performance bias towards the dominant class. In this paper, we present a novel GeoAI approach to training sea ice classification by formalizing it as a partial label learning task with explicit confidence scores to address multiple labels and class imbalance. We treat the polygon-level labels as candidate partial labels, assign the corresponding ice concentrations as confidence scores to each candidate label, and integrate them with focal loss to train a Convolutional Neural Network (CNN). Our proposed approach leads to enhanced performance for sea ice classification in Sentinel-1 dual-polarized SAR images, improving classification accuracy (from 87% to 92%) and weighted average F-1 score (from 90% to 93%) compared to the conventional training approach of using one-hot encoded labels and Categorical Cross-Entropy loss. It also improves the F-1 score in 4 out of the 6 sea ice classes.
Authors: Xinyu Liu, Jing Zhang, Kexin Zhang, Yuting Yang, Licheng Jiao, Shuyuan Yang
Abstract: Video Object Segmentation (VOS) is a vital task in computer vision, focusing on distinguishing foreground objects from the background across video frames. Our work draws inspiration from the Cutie model, and we investigate the effects of object memory, the total number of memory frames, and input resolution on segmentation performance. This report validates the effectiveness of our inference method on the coMplex video Object SEgmentation (MOSE) dataset, which features complex occlusions. Our experimental results demonstrate that our approach achieves a J\&F score of 0.8139 on the test set, securing the third position in the final ranking. These findings highlight the robustness and accuracy of our method in handling challenging VOS scenarios.
Authors: Xiang Xu, Tianchen Zhao, Zheng Zhang, Zhihua Li, Jon Wu, Alessandro Achille, Mani Srivastava
Abstract: Protecting digital identities of human face from various attack vectors is paramount, and face anti-spoofing plays a crucial role in this endeavor. Current approaches primarily focus on detecting spoofing attempts within individual frames to detect presentation attacks. However, the emergence of hyper-realistic generative models capable of real-time operation has heightened the risk of digitally generated attacks. In light of these evolving threats, this paper aims to address two key aspects. First, it sheds light on the vulnerabilities of state-of-the-art face anti-spoofing methods against digital attacks. Second, it presents a comprehensive taxonomy of common threats encountered in face anti-spoofing systems. Through a series of experiments, we demonstrate the limitations of current face anti-spoofing detection techniques and their failure to generalize to novel digital attack scenarios. Notably, the existing models struggle with digital injection attacks including adversarial noise, realistic deepfake attacks, and digital replay attacks. To aid in the design and implementation of robust face anti-spoofing systems resilient to these emerging vulnerabilities, the paper proposes key design principles from model accuracy and robustness to pipeline robustness and even platform robustness. Especially, we suggest to implement the proactive face anti-spoofing system using active sensors to significant reduce the risks for unseen attack vectors and improve the user experience.
Authors: Mengyu Zhao, Xi Chen, Xin Yuan, Shirin Jalali
Abstract: Snapshot compressive imaging (SCI) recovers high-dimensional (3D) data cubes from a single 2D measurement, enabling diverse applications like video and hyperspectral imaging to go beyond standard techniques in terms of acquisition speed and efficiency. In this paper, we focus on SCI recovery algorithms that employ untrained neural networks (UNNs), such as deep image prior (DIP), to model source structure. Such UNN-based methods are appealing as they have the potential of avoiding the computationally intensive retraining required for different source models and different measurement scenarios. We first develop a theoretical framework for characterizing the performance of such UNN-based methods. The theoretical framework, on the one hand, enables us to optimize the parameters of data-modulating masks, and on the other hand, provides a fundamental connection between the number of data frames that can be recovered from a single measurement to the parameters of the untrained NN. We also employ the recently proposed bagged-deep-image-prior (bagged-DIP) idea to develop SCI Bagged Deep Video Prior (SCI-BDVP) algorithms that address the common challenges faced by standard UNN solutions. Our experimental results show that in video SCI our proposed solution achieves state-of-the-art among UNN methods, and in the case of noisy measurements, it even outperforms supervised solutions.
Authors: Diwen Wan, Ruijie Lu, Gang Zeng
Abstract: Rendering novel view images in dynamic scenes is a crucial yet challenging task. Current methods mainly utilize NeRF-based methods to represent the static scene and an additional time-variant MLP to model scene deformations, resulting in relatively low rendering quality as well as slow inference speed. To tackle these challenges, we propose a novel framework named Superpoint Gaussian Splatting (SP-GS). Specifically, our framework first employs explicit 3D Gaussians to reconstruct the scene and then clusters Gaussians with similar properties (e.g., rotation, translation, and location) into superpoints. Empowered by these superpoints, our method manages to extend 3D Gaussian splatting to dynamic scenes with only a slight increase in computational expense. Apart from achieving state-of-the-art visual quality and real-time rendering under high resolutions, the superpoint representation provides a stronger manipulation capability. Extensive experiments demonstrate the practicality and effectiveness of our approach on both synthetic and real-world datasets. Please see our project page at https://dnvtmf.github.io/SP_GS.github.io.
Authors: Zilu Guo, Liuyang Bian, Xuan Huang, Hu Wei, Jingyu Li, Huasheng Ni
Abstract: Atrous convolutions are employed as a method to increase the receptive field in semantic segmentation tasks. However, in previous works of semantic segmentation, it was rarely employed in the shallow layers of the model. We revisit the design of atrous convolutions in modern convolutional neural networks (CNNs), and demonstrate that the concept of using large kernels to apply atrous convolutions could be a more powerful paradigm. We propose three guidelines to apply atrous convolutions more efficiently. Following these guidelines, we propose DSNet, a Dual-Branch CNN architecture, which incorporates atrous convolutions in the shallow layers of the model architecture, as well as pretraining the nearly entire encoder on ImageNet to achieve better performance. To demonstrate the effectiveness of our approach, our models achieve a new state-of-the-art trade-off between accuracy and speed on ADE20K, Cityscapes and BDD datasets. Specifically, DSNet achieves 40.0% mIOU with inference speed of 179.2 FPS on ADE20K, and 80.4% mIOU with speed of 81.9 FPS on Cityscapes. Source code and models are available at Github: https://github.com/takaniwa/DSNet.
Authors: Minzhou Pan, Yi Zeng, Xue Lin, Ning Yu, Cho-Jui Hsieh, Peter Henderson, Ruoxi Jia
Abstract: In this study, we investigate the vulnerability of image watermarks to diffusion-model-based image editing, a challenge exacerbated by the computational cost of accessing gradient information and the closed-source nature of many diffusion models. To address this issue, we introduce JIGMARK. This first-of-its-kind watermarking technique enhances robustness through contrastive learning with pairs of images, processed and unprocessed by diffusion models, without needing a direct backpropagation of the diffusion process. Our evaluation reveals that JIGMARK significantly surpasses existing watermarking solutions in resilience to diffusion-model edits, demonstrating a True Positive Rate more than triple that of leading baselines at a 1% False Positive Rate while preserving image quality. At the same time, it consistently improves the robustness against other conventional perturbations (like JPEG, blurring, etc.) and malicious watermark attacks over the state-of-the-art, often by a large margin. Furthermore, we propose the Human Aligned Variation (HAV) score, a new metric that surpasses traditional similarity measures in quantifying the number of image derivatives from image editing.
Authors: Xin Wang, Fangfang Liu, Zheng Li, Caili Guo
Abstract: Text attribute person search aims to find specific pedestrians through given textual attributes, which is very meaningful in the scene of searching for designated pedestrians through witness descriptions. The key challenge is the significant modality gap between textual attributes and images. Previous methods focused on achieving explicit representation and alignment through unimodal pre-trained models. Nevertheless, the absence of inter-modality correspondence in these models may lead to distortions in the local information of intra-modality. Moreover, these methods only considered the alignment of inter-modality and ignored the differences between different attribute categories. To mitigate the above problems, we propose an Attribute-Aware Implicit Modality Alignment (AIMA) framework to learn the correspondence of local representations between textual attributes and images and combine global representation matching to narrow the modality gap. Firstly, we introduce the CLIP model as the backbone and design prompt templates to transform attribute combinations into structured sentences. This facilitates the model's ability to better understand and match image details. Next, we design a Masked Attribute Prediction (MAP) module that predicts the masked attributes after the interaction of image and masked textual attribute features through multi-modal interaction, thereby achieving implicit local relationship alignment. Finally, we propose an Attribute-IoU Guided Intra-Modal Contrastive (A-IoU IMC) loss, aligning the distribution of different textual attributes in the embedding space with their IoU distribution, achieving better semantic arrangement. Extensive experiments on the Market-1501 Attribute, PETA, and PA100K datasets show that the performance of our proposed method significantly surpasses the current state-of-the-art methods.
Authors: Xinhang Liu, Yu-Wing Tai, Chi-Keung Tang, Pedro Miraldo, Suhas Lohit, Moitreya Chatterjee
Abstract: Extensions of Neural Radiance Fields (NeRFs) to model dynamic scenes have enabled their near photo-realistic, free-viewpoint rendering. Although these methods have shown some potential in creating immersive experiences, two drawbacks limit their ubiquity: (i) a significant reduction in reconstruction quality when the computing budget is limited, and (ii) a lack of semantic understanding of the underlying scenes. To address these issues, we introduce Gear-NeRF, which leverages semantic information from powerful image segmentation models. Our approach presents a principled way for learning a spatio-temporal (4D) semantic embedding, based on which we introduce the concept of gears to allow for stratified modeling of dynamic regions of the scene based on the extent of their motion. Such differentiation allows us to adjust the spatio-temporal sampling resolution for each region in proportion to its motion scale, achieving more photo-realistic dynamic novel view synthesis. At the same time, almost for free, our approach enables free-viewpoint tracking of objects of interest - a functionality not yet achieved by existing NeRF-based methods. Empirical studies validate the effectiveness of our method, where we achieve state-of-the-art rendering and tracking performance on multiple challenging datasets.
Authors: Jielin Qiu, William Han, Xuandong Zhao, Shangbang Long, Christos Faloutsos, Lei Li
Abstract: With the development of large models, watermarks are increasingly employed to assert copyright, verify authenticity, or monitor content distribution. As applications become more multimodal, the utility of watermarking techniques becomes even more critical. The effectiveness and reliability of these watermarks largely depend on their robustness to various disturbances. However, the robustness of these watermarks in real-world scenarios, particularly under perturbations and corruption, is not well understood. To highlight the significance of robustness in watermarking techniques, our study evaluated the robustness of watermarked content generated by image and text generation models against common real-world image corruptions and text perturbations. Our results could pave the way for the development of more robust watermarking techniques in the future. Our project website can be found at \url{https://mmwatermark-robustness.github.io/}.
Authors: Fang Chen, Gourav Datta, Mujahid Al Rafi, Hyeran Jeon, Meng Tang
Abstract: The expansion of neural network sizes and the enhancement of image resolution through modern camera sensors result in heightened memory and power demands for neural networks. Reducing peak memory, which is the maximum memory consumed during the execution of a neural network, is critical to deploy neural networks on edge devices with limited memory budget. A naive approach to reducing peak memory is aggressive down-sampling of feature maps via pooling with large stride, which often results in unacceptable degradation in network performance. To mitigate this problem, we propose residual encoded distillation (ReDistill) for peak memory reduction in a teacher-student framework, in which a student network with less memory is derived from the teacher network using aggressive pooling. We apply our distillation method to multiple problems in computer vision including image classification and diffusion based image generation. For image classification, our method yields 2x-3.2x measured peak memory on an edge GPU with negligible degradation in accuracy for most CNN based architectures. Additionally, our method yields improved test accuracy for tiny vision transformer (ViT) based models distilled from large CNN based teacher architectures. For diffusion-based image generation, our proposed distillation method yields a denoising network with 4x lower theoretical peak memory while maintaining decent diversity and fidelity for image generation. Experiments demonstrate our method's superior performance compared to other feature-based and response-based distillation methods.
Authors: Devichand Budagam, Ayush Kumar, Sayan Ghosh, Anuj Shrivastav, Azamat Zhanatuly Imanbayev, Iskander Rafailovich Akhmetov, Dmitrii Kaplun, Sergey Antonov, Artem Rychenkov, Gleb Cyganov, Aleksandr Sinitca
Abstract: Teeth segmentation and recognition are critical in various dental applications and dental diagnosis. Automatic and accurate segmentation approaches have been made possible by integrating deep learning models. Although teeth segmentation has been studied in the past, only some techniques were able to effectively classify and segment teeth simultaneously. This article offers a pipeline of two deep learning models, U-Net and YOLOv8, which results in BB-UNet, a new architecture for the classification and segmentation of teeth on panoramic X-rays that is efficient and reliable. We have improved the quality and reliability of teeth segmentation by utilising the YOLOv8 and U-Net capabilities. The proposed networks have been evaluated using the mean average precision (mAP) and dice coefficient for YOLOv8 and BB-UNet, respectively. We have achieved a 3\% increase in mAP score for teeth classification compared to existing methods, and a 10-15\% increase in dice coefficient for teeth segmentation compared to U-Net across different categories of teeth. A new Dental dataset was created based on UFBA-UESC dataset with Bounding-Box and Polygon annotations of 425 dental panoramic X-rays. The findings of this research pave the way for a wider adoption of object detection models in the field of dental diagnosis.
Authors: Nan Zhang, Xidan Zhang, Jianing Wei, Fangjun Wang, Zhiming Tan
Abstract: This report describes the winning solution to the WeatherProof Dataset Challenge (CVPR 2024 UG2+ Track 3). Details regarding the challenge are available at https://cvpr2024ug2challenge.github.io/track3.html. We propose an enhanced semantic segmentation pipeline for this challenge. Firstly, we improve semantic segmentation models, using backbone pretrained with Depth Anything to improve UperNet model and SETRMLA model, and adding language guidance based on both weather and category information to InternImage model. Secondly, we introduce a new dataset WeatherProofExtra with wider viewing angle and employ data augmentation methods, including adverse weather and super-resolution. Finally, effective training strategies and ensemble method are applied to improve final performance further. Our solution is ranked 1st on the final leaderboard. Code will be available at https://github.com/KaneiGi/WeatherProofChallenge.
URLs: https://cvpr2024ug2challenge.github.io/track3.html., https://github.com/KaneiGi/WeatherProofChallenge.
Authors: Yongtuo Liu, Sara Magliacane, Miltiadis Kofinas, Efstratios Gavves
Abstract: Hybrid dynamical systems are prevalent in science and engineering to express complex systems with continuous and discrete states. To learn the laws of systems, all previous methods for equation discovery in hybrid systems follow a two-stage paradigm, i.e. they first group time series into small cluster fragments and then discover equations in each fragment separately through methods in non-hybrid systems. Although effective, these methods do not fully take advantage of the commonalities in the shared dynamics of multiple fragments that are driven by the same equations. Besides, the two-stage paradigm breaks the interdependence between categorizing and representing dynamics that jointly form hybrid systems. In this paper, we reformulate the problem and propose an end-to-end learning framework, i.e. Amortized Equation Discovery (AMORE), to jointly categorize modes and discover equations characterizing the dynamics of each mode by all segments of the mode. Experiments on four hybrid and six non-hybrid systems show that our method outperforms previous methods on equation discovery, segmentation, and forecasting.
Authors: Jixiang Wan, Xudong Zhang, Shuzhou Dong, Yuwei Zhang, Yuchen Yang, Ruoxi Wu, Ye Jiang, Jijunnan Li, Jinquan Lin, Ming Yang
Abstract: Accurate and robust localization remains a significant challenge for autonomous vehicles. The cost of sensors and limitations in local computational efficiency make it difficult to scale to large commercial applications. Traditional vision-based approaches focus on texture features that are susceptible to changes in lighting, season, perspective, and appearance. Additionally, the large storage size of maps with descriptors and complex optimization processes hinder system performance. To balance efficiency and accuracy, we propose a novel lightweight visual semantic localization algorithm that employs stable semantic features instead of low-level texture features. First, semantic maps are constructed offline by detecting semantic objects, such as ground markers, lane lines, and poles, using cameras or LiDAR sensors. Then, online visual localization is performed through data association of semantic features and map objects. We evaluated our proposed localization framework in the publicly available KAIST Urban dataset and in scenarios recorded by ourselves. The experimental results demonstrate that our method is a reliable and practical localization solution in various autonomous driving localization tasks.
Authors: Miguel A. Ferrer, Josep A. Calduch-Giner, Moises D\'iaz, Javier Sosa, Enrique Rosell-Moll, Judith Santana Abril, Graciela Santana Sosa, Tom\'as Bautista Delgado, Cristina Carmona, Juan Antonio Martos-Sitcha, Enric Cabruja, Juan Manuel Afonso, Aurelio Vega, Manuel Lozano, Juan Antonio Montiel-Nelson, Jaume P\'erez-S\'anchez
Abstract: The AEFishBIT tri-axial accelerometer was externally attached to the operculum to assess the divergent activity and respiratory patterns of two marine farmed fish, the gilthead sea bream (Sparus aurata) and European sea bass (Dicentrarchus labrax). Analysis of raw data from exercised fish highlighted the large amplitude of operculum aperture and body tail movements in European sea bass, which were overall more stable at low-medium exercise intensity levels. Cosinor analysis in free-swimming fish (on-board data processing) highlighted a pronounced daily rhythmicity of locomotor activity and respiratory frequency in both gilthead sea bream and European sea bass. Acrophases of activity and respiration were coupled in gilthead sea bream, acting feeding time (once daily at 11:00 h) as a main synchronizing factor. By contrast, locomotor activity and respiratory frequency were out of phase in European sea bass with activity acrophase on early morning and respiration acrophase on the afternoon. The daily range of activity and respiration variation was also higher in European sea bass, probably as part of the adaptation of this fish species to act as a fast swimming predator. In any case, lower locomotor activity and enhanced respiration were associated with larger body weight in both fish species. This agrees with the notion that selection for fast growth in farming conditions is accompanied by a lower activity profile, which may favor an efficient feed conversion for growth purposes. Therefore, the use of behavioral monitoring is becoming a reliable and large-scale promising tool for selecting more efficient farmed fish, allowing researchers and farmers to establish stricter criteria of welfare for more sustainable and ethical fish production.
Authors: Senran Fan, Zhicheng Bao, Chen Dong, Haotai Liang, Xiaodong Xu, Ping Zhang
Abstract: Semantic communication, as a revolutionary communication architecture, is considered a promising novel communication paradigm. Unlike traditional symbol-based error-free communication systems, semantic-based visual communication systems extract, compress, transmit, and reconstruct images at the semantic level. However, widely used image similarity evaluation metrics, whether pixel-based MSE or PSNR or structure-based MS-SSIM, struggle to accurately measure the loss of semantic-level information of the source during system transmission. This presents challenges in evaluating the performance of visual semantic communication systems, especially when comparing them with traditional communication systems. To address this, we propose a semantic evaluation metric -- SeSS (Semantic Similarity Score), based on Scene Graph Generation and graph matching, which shifts the similarity scores between images into semantic-level graph matching scores. Meanwhile, semantic similarity scores for tens of thousands of image pairs are manually annotated to fine-tune the hyperparameters in the graph matching algorithm, aligning the metric more closely with human semantic perception. The performance of the SeSS is tested on different datasets, including (1)images transmitted by traditional and semantic communication systems at different compression rates, (2)images transmitted by traditional and semantic communication systems at different signal-to-noise ratios, (3)images generated by large-scale model with different noise levels introduced, and (4)cases of images subjected to certain special transformations. The experiments demonstrate the effectiveness of SeSS, indicating that the metric can measure the semantic-level differences in semantic-level information of images and can be used for evaluation in visual semantic communication systems.
Authors: Yixuan Yang, Junru Lu, Zixiang Zhao, Zhen Luo, James J. Q. Yu, Victor Sanchez, Feng Zheng
Abstract: Designing 3D indoor layouts is a crucial task with significant applications in virtual reality, interior design, and automated space planning. Existing methods for 3D layout design either rely on diffusion models, which utilize spatial relationship priors, or heavily leverage the inferential capabilities of proprietary Large Language Models (LLMs), which require extensive prompt engineering and in-context exemplars via black-box trials. These methods often face limitations in generalization and dynamic scene editing. In this paper, we introduce LLplace, a novel 3D indoor scene layout designer based on lightweight fine-tuned open-source LLM Llama3. LLplace circumvents the need for spatial relationship priors and in-context exemplars, enabling efficient and credible room layout generation based solely on user inputs specifying the room type and desired objects. We curated a new dialogue dataset based on the 3D-Front dataset, expanding the original data volume and incorporating dialogue data for adding and removing objects. This dataset can enhance the LLM's spatial understanding. Furthermore, through dialogue, LLplace activates the LLM's capability to understand 3D layouts and perform dynamic scene editing, enabling the addition and removal of objects. Our approach demonstrates that LLplace can effectively generate and edit 3D indoor layouts interactively and outperform existing methods in delivering high-quality 3D design solutions. Code and dataset will be released.
Authors: Anshul Gupta, Pierre Vuillecard, Arya Farkhondeh, Jean-Marc Odobez
Abstract: Contextual cues related to a person's pose and interactions with objects and other people in the scene can provide valuable information for gaze following. While existing methods have focused on dedicated cue extraction methods, in this work we investigate the zero-shot capabilities of Vision-Language Models (VLMs) for extracting a wide array of contextual cues to improve gaze following performance. We first evaluate various VLMs, prompting strategies, and in-context learning (ICL) techniques for zero-shot cue recognition performance. We then use these insights to extract contextual cues for gaze following, and investigate their impact when incorporated into a state of the art model for the task. Our analysis indicates that BLIP-2 is the overall top performing VLM and that ICL can improve performance. We also observe that VLMs are sensitive to the choice of the text prompt although ensembling over multiple text prompts can provide more robust performance. Additionally, we discover that using the entire image along with an ellipse drawn around the target person is the most effective strategy for visual prompting. For gaze following, incorporating the extracted cues results in better generalization performance, especially when considering a larger set of cues, highlighting the potential of this approach.
Authors: Shan Li, Lu Yang, Pu Cao, Liulei Li, Huadong Ma
Abstract: The successful application of semantic segmentation technology in the real world has been among the most exciting achievements in the computer vision community over the past decade. Although the long-tailed phenomenon has been investigated in many fields, e.g., classification and object detection, it has not received enough attention in semantic segmentation and has become a non-negligible obstacle to applying semantic segmentation technology in autonomous driving and virtual reality. Therefore, in this work, we focus on a relatively under-explored task setting, long-tailed semantic segmentation (LTSS). We first establish three representative datasets from different aspects, i.e., scene, object, and human. We further propose a dual-metric evaluation system and construct the LTSS benchmark to demonstrate the performance of semantic segmentation methods and long-tailed solutions. We also propose a transformer-based algorithm to improve LTSS, frequency-based matcher, which solves the oversuppression problem by one-to-many matching and automatically determines the number of matching queries for each class. Given the comprehensiveness of this work and the importance of the issues revealed, this work aims to promote the empirical study of semantic segmentation tasks. Our datasets, codes, and models will be publicly available.
Authors: Sofija Engelson, Jan Ehrhardt, Timo Kepp, Joshua Niemeijer, Heinz Handels
Abstract: The evaluation of lymph node metastases plays a crucial role in achieving precise cancer staging, influencing subsequent decisions regarding treatment options. Lymph node detection poses challenges due to the presence of unclear boundaries and the diverse range of sizes and morphological characteristics, making it a resource-intensive process. As part of the LNQ 2023 MICCAI challenge, we propose the use of anatomical priors as a tool to address the challenges that persist in mediastinal lymph node segmentation in combination with the partial annotation of the challenge training data. The model ensemble using all suggested modifications yields a Dice score of 0.6033 and segments 57% of the ground truth lymph nodes, compared to 27% when training on CT only. Segmentation accuracy is improved significantly by incorporating a probabilistic lymph node atlas in loss weighting and post-processing. The largest performance gains are achieved by oversampling fully annotated data to account for the partial annotation of the challenge training data, as well as adding additional data augmentation to address the high heterogeneity of the CT images and lymph node appearance. Our code is available at https://github.com/MICAI-IMI-UzL/LNQ2023.
Authors: Ruipu Wu, Jifei Che, Han Li, Chengjing Wu, Ting Liu, Luoqi Liu
Abstract: Video panoptic segmentation is an advanced task that extends panoptic segmentation by applying its concept to video sequences. In the hope of addressing the challenge of video panoptic segmentation in diverse conditions, We utilize DVIS++ as our baseline model and enhance it by introducing a comprehensive approach centered on the query-wise ensemble, supplemented by additional techniques. Our proposed approach achieved a VPQ score of 57.01 on the VIPSeg test set, and ranked 3rd in the VPS track of the 3rd Pixel-level Video Understanding in the Wild Challenge.
Authors: Zonghao Ying, Aishan Liu, Tianyuan Zhang, Zhengmin Yu, Siyuan Liang, Xianglong Liu, Dacheng Tao
Abstract: In the realm of large vision language models (LVLMs), jailbreak attacks serve as a red-teaming approach to bypass guardrails and uncover safety implications. Existing jailbreaks predominantly focus on the visual modality, perturbing solely visual inputs in the prompt for attacks. However, they fall short when confronted with aligned models that fuse visual and textual features simultaneously for generation. To address this limitation, this paper introduces the Bi-Modal Adversarial Prompt Attack (BAP), which executes jailbreaks by optimizing textual and visual prompts cohesively. Initially, we adversarially embed universally harmful perturbations in an image, guided by a few-shot query-agnostic corpus (e.g., affirmative prefixes and negative inhibitions). This process ensures that image prompt LVLMs to respond positively to any harmful queries. Subsequently, leveraging the adversarial image, we optimize textual prompts with specific harmful intent. In particular, we utilize a large language model to analyze jailbreak failures and employ chain-of-thought reasoning to refine textual prompts through a feedback-iteration manner. To validate the efficacy of our approach, we conducted extensive evaluations on various datasets and LVLMs, demonstrating that our method significantly outperforms other methods by large margins (+29.03% in attack success rate on average). Additionally, we showcase the potential of our attacks on black-box commercial LVLMs, such as Gemini and ChatGLM.
Authors: Marianna Ohanyan, Hayk Manukyan, Zhangyang Wang, Shant Navasardyan, Humphrey Shi
Abstract: We present Zero-Painter, a novel training-free framework for layout-conditional text-to-image synthesis that facilitates the creation of detailed and controlled imagery from textual prompts. Our method utilizes object masks and individual descriptions, coupled with a global text prompt, to generate images with high fidelity. Zero-Painter employs a two-stage process involving our novel Prompt-Adjusted Cross-Attention (PACA) and Region-Grouped Cross-Attention (ReGCA) blocks, ensuring precise alignment of generated objects with textual prompts and mask shapes. Our extensive experiments demonstrate that Zero-Painter surpasses current state-of-the-art methods in preserving textual details and adhering to mask shapes.
Authors: Danielle Kapon, Michael Fire, Shai Gordin
Abstract: Cuneiform tablets, emerging in ancient Mesopotamia around the late fourth millennium BCE, represent one of humanity's earliest writing systems. Characterized by wedge-shaped marks on clay tablets, these artifacts provided insight into Mesopotamian civilization across various domains. Traditionally, the analysis and dating of these tablets rely on subjective assessment of shape and writing style, leading to uncertainties in pinpointing their exact temporal origins. Recent advances in digitization have revolutionized the study of cuneiform by enhancing accessibility and analytical capabilities. Our research uniquely focuses on the silhouette of tablets as significant indicators of their historical periods, diverging from most studies that concentrate on textual content. Utilizing an unprecedented dataset of over 94,000 images from the Cuneiform Digital Library Initiative collection, we apply deep learning methods to classify cuneiform tablets, covering over 3,000 years of history. By leveraging statistical, computational techniques, and generative modeling through Variational Auto-Encoders (VAEs), we achieve substantial advancements in the automatic classification of these ancient documents, focusing on the tablets' silhouettes as key predictors. Our classification approach begins with a Decision Tree using height-to-width ratios and culminates with a ResNet50 model, achieving a 61% macro F1-score for tablet silhouettes. Moreover, we introduce novel VAE-powered tools to enhance explainability and enable researchers to explore changes in tablet shapes across different eras and genres. This research contributes to document analysis and diplomatics by demonstrating the value of large-scale data analysis combined with statistical methods. These insights offer valuable tools for historians and epigraphists, enriching our understanding of cuneiform tablets and the cultures that produced them.
Authors: Thomas H. Schmitt, Maximilian Bundscherer, Tobias Bocklet
Abstract: The Semmeldetector, is a machine learning application that utilizes object detection models to detect, classify and count baked goods in images. Our application allows commercial bakers to track unsold baked goods, which allows them to optimize production and increase resource efficiency. We compiled a dataset comprising 1151 images that distinguishes between 18 different types of baked goods to train our detection models. To facilitate model training, we used a Copy-Paste augmentation pipeline to expand our dataset. We trained the state-of-the-art object detection model YOLOv8 on our detection task. We tested the impact of different training data, model scale, and online image augmentation pipelines on model performance. Our overall best performing model, achieved an AP@0.5 of 89.1% on our test set. Based on our results, we conclude that machine learning can be a valuable tool even for unforeseen industries like bakeries, even with very limited datasets.
Authors: Zhongliang Jiang, Yunfeng Kang, Yuan Bi, Xuesong Li, Chenyang Li, Nassir Navab
Abstract: Ultrasound imaging has been widely used in clinical examinations owing to the advantages of being portable, real-time, and radiation-free. Considering the potential of extensive deployment of autonomous examination systems in hospitals, robotic US imaging has attracted increased attention. However, due to the inter-patient variations, it is still challenging to have an optimal path for each patient, particularly for thoracic applications with limited acoustic windows, e.g., intercostal liver imaging. To address this problem, a class-aware cartilage bone segmentation network with geometry-constraint post-processing is presented to capture patient-specific rib skeletons. Then, a dense skeleton graph-based non-rigid registration is presented to map the intercostal scanning path from a generic template to individual patients. By explicitly considering the high-acoustic impedance bone structures, the transferred scanning path can be precisely located in the intercostal space, enhancing the visibility of internal organs by reducing the acoustic shadow. To evaluate the proposed approach, the final path mapping performance is validated on five distinct CTs and two volunteer US data, resulting in ten pairs of CT-US combinations. Results demonstrate that the proposed graph-based registration method can robustly and precisely map the path from CT template to individual patients (Euclidean error: $2.21\pm1.11~mm$).
Authors: Yihang Chen, Qianyi Wu, Mehrtash Harandi, Jianfei Cai
Abstract: In recent years, Neural Radiance Field (NeRF) has demonstrated remarkable capabilities in representing 3D scenes. To expedite the rendering process, learnable explicit representations have been introduced for combination with implicit NeRF representation, which however results in a large storage space requirement. In this paper, we introduce the Context-based NeRF Compression (CNC) framework, which leverages highly efficient context models to provide a storage-friendly NeRF representation. Specifically, we excavate both level-wise and dimension-wise context dependencies to enable probability prediction for information entropy reduction. Additionally, we exploit hash collision and occupancy grids as strong prior knowledge for better context modeling. To the best of our knowledge, we are the first to construct and exploit context models for NeRF compression. We achieve a size reduction of 100$\times$ and 70$\times$ with improved fidelity against the baseline Instant-NGP on Synthesic-NeRF and Tanks and Temples datasets, respectively. Additionally, we attain 86.7\% and 82.3\% storage size reduction against the SOTA NeRF compression method BiRF. Our code is available here: https://github.com/YihangChen-ee/CNC.
Authors: Jie Zhao, Zhitong Xiong, Xiao Xiang Zhu
Abstract: Due to its cloud-penetrating capability and independence from solar illumination, satellite Synthetic Aperture Radar (SAR) is the preferred data source for large-scale flood mapping, providing global coverage and including various land cover classes. However, most studies on large-scale SAR-derived flood mapping using deep learning algorithms have primarily focused on flooded open areas, utilizing available open-access datasets (e.g., Sen1Floods11) and with limited attention to urban floods. To address this gap, we introduce \textbf{UrbanSARFloods}, a floodwater dataset featuring pre-processed Sentinel-1 intensity data and interferometric coherence imagery acquired before and during flood events. It contains 8,879 $512\times 512$ chips covering 807,500 $km^2$ across 20 land cover classes and 5 continents, spanning 18 flood events. We used UrbanSARFloods to benchmark existing state-of-the-art convolutional neural networks (CNNs) for segmenting open and urban flood areas. Our findings indicate that prevalent approaches, including the Weighted Cross-Entropy (WCE) loss and the application of transfer learning with pretrained models, fall short in overcoming the obstacles posed by imbalanced data and the constraints of a small training dataset. Urban flood detection remains challenging. Future research should explore strategies for addressing imbalanced data challenges and investigate transfer learning's potential for SAR-based large-scale flood mapping. Besides, expanding this dataset to include additional flood events holds promise for enhancing its utility and contributing to advancements in flood mapping techniques.
Authors: Wei Chen, Yuxue Ren, Na Lei, Zhongxuan Luo, Xianfeng Gu
Abstract: Texture mapping is a common technology in the area of computer graphics, it maps the 3D surface space onto the 2D texture space. However, the loose texture space will reduce the efficiency of data storage and GPU memory addressing in the rendering process. Many of the existing methods focus on repacking given textures, but they still suffer from high computational cost and hardly produce a wholly tight texture space. In this paper, we propose a method to optimize the texture space and produce a new texture mapping which is compact based on global parameterization. The proposed method is computationally robust and efficient. Experiments show the effectiveness of the proposed method and the potency in improving the storage and rendering efficiency.
Authors: Xin Cai, Hailong Zhang, Chenchen Wang, Wentao Liu, Jinwei Gu, Tianfan Xue
Abstract: Lensless cameras, innovatively replacing traditional lenses for ultra-thin, flat optics, encode light directly onto sensors, producing images that are not immediately recognizable. This compact, lightweight, and cost-effective imaging solution offers inherent privacy advantages, making it attractive for privacy-sensitive applications like face verification. Typical lensless face verification adopts a two-stage process of reconstruction followed by verification, incurring privacy risks from reconstructed faces and high computational costs. This paper presents an end-to-end optimization approach for privacy-preserving face verification directly on encoded lensless captures, ensuring that the entire software pipeline remains encoded with no visible faces as intermediate results. To achieve this, we propose several techniques to address unique challenges from the lensless setup which precludes traditional face detection and alignment. Specifically, we propose a face center alignment scheme, an augmentation curriculum to build robustness against variations, and a knowledge distillation method to smooth optimization and enhance performance. Evaluations under both simulation and real environment demonstrate our method outperforms two-stage lensless verification while enhancing privacy and efficiency. Project website: \url{lenslessface.github.io}.
Authors: Drew Linsley, Peisen Zhou, Alekh Karkada Ashok, Akash Nagaraj, Gaurav Gaonkar, Francis E Lewis, Zygmunt Pizlo, Thomas Serre
Abstract: Visual perspective taking (VPT) is the ability to perceive and reason about the perspectives of others. It is an essential feature of human intelligence, which develops over the first decade of life and requires an ability to process the 3D structure of visual scenes. A growing number of reports have indicated that deep neural networks (DNNs) become capable of analyzing 3D scenes after training on large image datasets. We investigated if this emergent ability for 3D analysis in DNNs is sufficient for VPT with the 3D perception challenge (3D-PC): a novel benchmark for 3D perception in humans and DNNs. The 3D-PC is comprised of three 3D-analysis tasks posed within natural scene images: 1. a simple test of object depth order, 2. a basic VPT task (VPT-basic), and 3. another version of VPT (VPT-Strategy) designed to limit the effectiveness of "shortcut" visual strategies. We tested human participants (N=33) and linearly probed or text-prompted over 300 DNNs on the challenge and found that nearly all of the DNNs approached or exceeded human accuracy in analyzing object depth order. Surprisingly, DNN accuracy on this task correlated with their object recognition performance. In contrast, there was an extraordinary gap between DNNs and humans on VPT-basic. Humans were nearly perfect, whereas most DNNs were near chance. Fine-tuning DNNs on VPT-basic brought them close to human performance, but they, unlike humans, dropped back to chance when tested on VPT-perturb. Our challenge demonstrates that the training routines and architectures of today's DNNs are well-suited for learning basic 3D properties of scenes and objects but are ill-suited for reasoning about these properties like humans do. We release our 3D-PC datasets and code to help bridge this gap in 3D perception between humans and machines.
Authors: Takuhiro Kaneko
Abstract: Geometry-agnostic system identification is a technique for identifying the geometry and physical properties of an object from video sequences without any geometric assumptions. Recently, physics-augmented continuum neural radiance fields (PAC-NeRF) has demonstrated promising results for this technique by utilizing a hybrid Eulerian-Lagrangian representation, in which the geometry is represented by the Eulerian grid representations of NeRF, the physics is described by a material point method (MPM), and they are connected via Lagrangian particles. However, a notable limitation of PAC-NeRF is that its performance is sensitive to the learning of the geometry from the first frames owing to its two-step optimization. First, the grid representations are optimized with the first frames of video sequences, and then the physical properties are optimized through video sequences utilizing the fixed first-frame grid representations. This limitation can be critical when learning of the geometric structure is difficult, for example, in a few-shot (sparse view) setting. To overcome this limitation, we propose Lagrangian particle optimization (LPO), in which the positions and features of particles are optimized through video sequences in Lagrangian space. This method allows for the optimization of the geometric structure across the entire video sequence within the physical constraints imposed by the MPM. The experimental results demonstrate that the LPO is useful for geometric correction and physical identification in sparse-view settings.
Authors: Da Li, Guoqiang Zhao, Houjun Sun, Jiacheng Bao
Abstract: Multi-baseline SAR 3D imaging faces significant challenges due to data sparsity. In recent years, deep learning techniques have achieved notable success in enhancing the quality of sparse SAR 3D imaging. However, previous work typically rely on full-aperture high-resolution radar images to supervise the training of deep neural networks (DNNs), utilizing only single-modal information from radar data. Consequently, imaging performance is limited, and acquiring full-aperture data for multi-baseline SAR is costly and sometimes impractical in real-world applications. In this paper, we propose a Cross-Modal Reconstruction Network (CMR-Net), which integrates differentiable render and cross-modal supervision with optical images to reconstruct highly sparse multi-baseline SAR 3D images of vehicle targets into visually structured and high-resolution images. We meticulously designed the network architecture and training strategies to enhance network generalization capability. Remarkably, CMR-Net, trained solely on simulated data, demonstrates high-resolution reconstruction capabilities on both publicly available simulation datasets and real measured datasets, outperforming traditional sparse reconstruction algorithms based on compressed sensing and other learning-based methods. Additionally, using optical images as supervision provides a cost-effective way to build training datasets, reducing the difficulty of method dissemination. Our work showcases the broad prospects of deep learning in multi-baseline SAR 3D imaging and offers a novel path for researching radar imaging based on cross-modal learning theory.
Authors: Mouad Klai, Olivier Monga, Mohamed Soufiane Jouini, Val\'erie Pot
Abstract: This study presents a new computational approach for simulating the microbial decomposition of organic matter, from 3D micro-computed tomography (micro-CT) images of soil. The method employs a valuated graph of connected voxels to simulate transformation and diffusion processes involved in microbial decomposition within the complex soil matrix. The resulting model can be adapted to simulate any diffusion-transformation processes in porous media. We implemented parallelization strategies and explored different numerical methods, including implicit, explicit, synchronous, and asynchronous schemes. To validate our method, we compared simulation outputs with those provided by LBioS and by Mosaic models. LBioS uses a lattice-Boltzmann method for diffusion and Mosaic takes benefit of Pore Network Geometrical Modelling (PNGM) by means of geometrical primitives such as spheres and ellipsoids. This approach achieved comparable results to traditional LBM-based simulations, but required only one-fourth of the computing time. Compared to Mosaic simulation, the proposed method is slower but more accurate and does not require any calibration. Furthermore, we present a theoretical framework and an application example to enhance PNGM-based simulations. This is accomplished by approximating the diffusional conductance coefficients using stochastic gradient descent and data generated by the current approach.
Authors: Zhicheng Cai, Qiu Shen
Abstract: Implicit neural representation (INR) has recently emerged as a promising paradigm for signal representations, which takes coordinates as inputs and generates corresponding signal values. Since these coordinates contain no semantic features, INR fails to take any semantic information into consideration. However, semantic information has been proven critical in many vision tasks, especially for visual signal representation. This paper proposes a reparameterization method termed as SPW, which encodes the semantic priors to the weights of INR, thus making INR contain semantic information implicitly and enhancing its representational capacity. Specifically, SPW uses the Semantic Neural Network (SNN) to extract both low- and high-level semantic information of the target visual signal and generates the semantic vector, which is input into the Weight Generation Network (WGN) to generate the weights of INR model. Finally, INR uses the generated weights with semantic priors to map the coordinates to the signal values. After training, we only retain the generated weights while abandoning both SNN and WGN, thus SPW introduces no extra costs in inference. Experimental results show that SPW can improve the performance of various INR models significantly on various tasks, including image fitting, CT reconstruction, MRI reconstruction, and novel view synthesis. Further experiments illustrate that model with SPW has lower weight redundancy and learns more novel representations, validating the effectiveness of SPW.
Authors: Nicolas Cherel, Andr\'es Almansa, Yann Gousseau, Alasdair Newson
Abstract: Diffusion models are now the undisputed state-of-the-art for image generation and image restoration. However, they require large amounts of computational power for training and inference. In this paper, we propose lightweight diffusion models for image inpainting that can be trained on a single image, or a few images. We show that our approach competes with large state-of-the-art models in specific cases. We also show that training a model on a single image is particularly relevant for image acquisition modality that differ from the RGB images of standard learning databases. We show results in three different contexts: texture images, line drawing images, and materials BRDF, for which we achieve state-of-the-art results in terms of realism, with a computational load that is greatly reduced compared to concurrent methods.
Authors: Haotian Zhang, Keyan Chen, Chenyang Liu, Hao Chen, Zhengxia Zou, Zhenwei Shi
Abstract: Recently, the Mamba architecture based on state space models has demonstrated remarkable performance in a series of natural language processing tasks and has been rapidly applied to remote sensing change detection (CD) tasks. However, most methods enhance the global receptive field by directly modifying the scanning mode of Mamba, neglecting the crucial role that local information plays in dense prediction tasks (e.g., CD). In this article, we propose a model called CDMamba, which effectively combines global and local features for handling CD tasks. Specifically, the Scaled Residual ConvMamba (SRCM) block is proposed to utilize the ability of Mamba to extract global features and convolution to enhance the local details, to alleviate the issue that current Mamba-based methods lack detailed clues and are difficult to achieve fine detection in dense prediction tasks. Furthermore, considering the characteristics of bi-temporal feature interaction required for CD, the Adaptive Global Local Guided Fusion (AGLGF) block is proposed to dynamically facilitate the bi-temporal interaction guided by other temporal global/local features. Our intuition is that more discriminative change features can be acquired with the guidance of other temporal features. Extensive experiments on three datasets demonstrate that our proposed CDMamba outperforms the current state-of-the-art methods. Our code will be open-sourced at https://github.com/zmoka-zht/CDMamba.
Authors: Siyuan Li, Lei Ke, Martin Danelljan, Luigi Piccinelli, Mattia Segu, Luc Van Gool, Fisher Yu
Abstract: The robust association of the same objects across video frames in complex scenes is crucial for many applications, especially Multiple Object Tracking (MOT). Current methods predominantly rely on labeled domain-specific video datasets, which limits the cross-domain generalization of learned similarity embeddings. We propose MASA, a novel method for robust instance association learning, capable of matching any objects within videos across diverse domains without tracking labels. Leveraging the rich object segmentation from the Segment Anything Model (SAM), MASA learns instance-level correspondence through exhaustive data transformations. We treat the SAM outputs as dense object region proposals and learn to match those regions from a vast image collection. We further design a universal MASA adapter which can work in tandem with foundational segmentation or detection models and enable them to track any detected objects. Those combinations present strong zero-shot tracking ability in complex domains. Extensive tests on multiple challenging MOT and MOTS benchmarks indicate that the proposed method, using only unlabeled static images, achieves even better performance than state-of-the-art methods trained with fully annotated in-domain video sequences, in zero-shot association. Project Page: https://matchinganything.github.io/
Authors: Matthew J Allen, Francisco Dorr, Joseph Alejandro Gallego Mejia, Laura Mart\'inez-Ferrer, Anna Jungbluth, Freddie Kalaitzis, Ra\'ul Ramos-Poll\'an
Abstract: Satellite-based remote sensing has revolutionised the way we address global challenges in a rapidly evolving world. Huge quantities of Earth Observation (EO) data are generated by satellite sensors daily, but processing these large datasets for use in ML pipelines is technically and computationally challenging. Specifically, different types of EO data are often hosted on a variety of platforms, with differing availability for Python preprocessing tools. In addition, spatial alignment across data sources and data tiling can present significant technical hurdles for novice users. While some preprocessed EO datasets exist, their content is often limited to optical or near-optical wavelength data, which is ineffective at night or in adverse weather conditions. Synthetic Aperture Radar (SAR), an active sensing technique based on microwave length radiation, offers a viable alternative. However, the application of machine learning to SAR has been limited due to a lack of ML-ready data and pipelines, particularly for the full diversity of SAR data, including polarimetry, coherence and interferometry. We introduce M3LEO, a multi-modal, multi-label EO dataset that includes polarimetric, interferometric, and coherence SAR data derived from Sentinel-1, alongside Sentinel-2 RGB imagery and a suite of labelled tasks for model evaluation. M3LEO spans 17.5TB and contains approximately 10M data chips across six geographic regions. The dataset is complemented by a flexible PyTorch Lightning framework, with configuration management using Hydra. We provide tools to process any dataset available on popular platforms such as Google Earth Engine for integration with our framework. Initial experiments validate the utility of our data and framework, showing that SAR imagery contains information additional to that extractable from RGB data. Data at huggingface.co/M3LEO, and code at github.com/spaceml-org/M3LEO.
Authors: Samyadeep Basu, Martin Grayson, Cecily Morrison, Besmira Nushi, Soheil Feizi, Daniela Massiceti
Abstract: Understanding the mechanisms of information storage and transfer in Transformer-based models is important for driving model understanding progress. Recent work has studied these mechanisms for Large Language Models (LLMs), revealing insights on how information is stored in a model's parameters and how information flows to and from these parameters in response to specific prompts. However, these studies have not yet been extended to Multi-modal Large Language Models (MLLMs). Given their expanding capabilities and real-world use, we start by studying one aspect of these models -- how MLLMs process information in a factual visual question answering task. We use a constraint-based formulation which views a visual question as having a set of visual or textual constraints that the model's generated answer must satisfy to be correct (e.g. What movie directed by the director in this photo has won a Golden Globe?). Under this setting, we contribute i) a method that extends causal information tracing from pure language to the multi-modal setting, and ii) VQA-Constraints, a test-bed of 9.7K visual questions annotated with constraints. We use these tools to study two open-source MLLMs, LLaVa and multi-modal Phi-2. Our key findings show that these MLLMs rely on MLP and self-attention blocks in much earlier layers for information storage, compared to LLMs whose mid-layer MLPs are more important. We also show that a consistent small subset of visual tokens output by the vision encoder are responsible for transferring information from the image to these causal blocks. We validate these mechanisms by introducing MultEdit, a model-editing algorithm that can correct errors and insert new long-tailed information into MLLMs by targeting these causal blocks.
Authors: Zhicheng Cai
Abstract: Implicit neural representation (INR) has recently emerged as a promising paradigm for signal representations. Typically, INR is parameterized by a multiplayer perceptron (MLP) which takes the coordinates as the inputs and generates corresponding attributes of a signal. However, MLP-based INRs face two critical issues: i) individually considering each coordinate while ignoring the connections; ii) suffering from the spectral bias thus failing to learn high-frequency components. While target visual signals usually exhibit strong local structures and neighborhood dependencies, and high-frequency components are significant in these signals, the issues harm the representational capacity of INRs. This paper proposes Conv-INR, the first INR model fully based on convolution. Due to the inherent attributes of convolution, Conv-INR can simultaneously consider adjacent coordinates and learn high-frequency components effectively. Compared to existing MLP-based INRs, Conv-INR has better representational capacity and trainability without requiring primary function expansion. We conduct extensive experiments on four tasks, including image fitting, CT/MRI reconstruction, and novel view synthesis, Conv-INR all significantly surpasses existing MLP-based INRs, validating the effectiveness. Finally, we raise three reparameterization methods that can further enhance the performance of the vanilla Conv-INR without introducing any extra inference cost.
Authors: Haosen Yang, Chenhao Zhang, Wenqing Wang, Marco Volino, Adrian Hilton, Li Zhang, Xiatian Zhu
Abstract: Point management is a critical component in optimizing 3D Gaussian Splatting (3DGS) models, as the point initiation (e.g., via structure from motion) is distributionally inappropriate. Typically, the Adaptive Density Control (ADC) algorithm is applied, leveraging view-averaged gradient magnitude thresholding for point densification, opacity thresholding for pruning, and regular all-points opacity reset. However, we reveal that this strategy is limited in tackling intricate/special image regions (e.g., transparent) as it is unable to identify all the 3D zones that require point densification, and lacking an appropriate mechanism to handle the ill-conditioned points with negative impacts (occlusion due to false high opacity). To address these limitations, we propose a Localized Point Management (LPM) strategy, capable of identifying those error-contributing zones in the highest demand for both point addition and geometry calibration. Zone identification is achieved by leveraging the underlying multiview geometry constraints, with the guidance of image rendering errors. We apply point densification in the identified zone, whilst resetting the opacity of those points residing in front of these regions so that a new opportunity is created to correct ill-conditioned points. Serving as a versatile plugin, LPM can be seamlessly integrated into existing 3D Gaussian Splatting models. Experimental evaluation across both static 3D and dynamic 4D scenes validate the efficacy of our LPM strategy in boosting a variety of existing 3DGS models both quantitatively and qualitatively. Notably, LPM improves both vanilla 3DGS and SpaceTimeGS to achieve state-of-the-art rendering quality while retaining real-time speeds, outperforming on challenging datasets such as Tanks & Temples and the Neural 3D Video Dataset.
Authors: Ruihe Wang, Yukang Cao, Kai Han, Kwan-Yee K. Wong
Abstract: 3D modeling has long been an important area in computer vision and computer graphics. Recently, thanks to the breakthroughs in neural representations and generative models, we witnessed a rapid development of 3D modeling. 3D human modeling, lying at the core of many real-world applications, such as gaming and animation, has attracted significant attention. Over the past few years, a large body of work on creating 3D human avatars has been introduced, forming a new and abundant knowledge base for 3D human modeling. The scale of the literature makes it difficult for individuals to keep track of all the works. This survey aims to provide a comprehensive overview of these emerging techniques for 3D human avatar modeling, from both reconstruction and generation perspectives. Firstly, we review representative methods for 3D human reconstruction, including methods based on pixel-aligned implicit function, neural radiance field, and 3D Gaussian Splatting, etc. We then summarize representative methods for 3D human generation, especially those using large language models like CLIP, diffusion models, and various 3D representations, which demonstrate state-of-the-art performance. Finally, we discuss our reflection on existing methods and open challenges for 3D human avatar modeling, shedding light on future research.
Authors: Salvatore Esposito, Qingshan Xu, Kacper Kania, Charlie Hewitt, Octave Mariotti, Lohit Petikam, Julien Valentin, Arno Onken, Oisin Mac Aodha
Abstract: We introduce a new generative approach for synthesizing 3D geometry and images from single-view collections. Most existing approaches predict volumetric density to render multi-view consistent images. By employing volumetric rendering using neural radiance fields, they inherit a key limitation: the generated geometry is noisy and unconstrained, limiting the quality and utility of the output meshes. To address this issue, we propose GeoGen, a new SDF-based 3D generative model trained in an end-to-end manner. Initially, we reinterpret the volumetric density as a Signed Distance Function (SDF). This allows us to introduce useful priors to generate valid meshes. However, those priors prevent the generative model from learning details, limiting the applicability of the method to real-world scenarios. To alleviate that problem, we make the transformation learnable and constrain the rendered depth map to be consistent with the zero-level set of the SDF. Through the lens of adversarial training, we encourage the network to produce higher fidelity details on the output meshes. For evaluation, we introduce a synthetic dataset of human avatars captured from 360-degree camera angles, to overcome the challenges presented by real-world datasets, which often lack 3D consistency and do not cover all camera angles. Our experiments on multiple datasets show that GeoGen produces visually and quantitatively better geometry than the previous generative models based on neural radiance fields.
Authors: Junjie Zhou, Yan Shu, Bo Zhao, Boya Wu, Shitao Xiao, Xi Yang, Yongping Xiong, Bo Zhang, Tiejun Huang, Zheng Liu
Abstract: The evaluation of Long Video Understanding (LVU) performance poses an important but challenging research problem. Despite previous efforts, the existing video understanding benchmarks are severely constrained by several issues, especially the insufficient lengths of videos, a lack of diversity in video types and evaluation tasks, and the inappropriateness for evaluating LVU performances. To address the above problems, we propose a new benchmark, called MLVU (Multi-task Long Video Understanding Benchmark), for the comprehensive and in-depth evaluation of LVU. MLVU presents the following critical values: 1) The substantial and flexible extension of video lengths, which enables the benchmark to evaluate LVU performance across a wide range of durations. 2) The inclusion of various video genres, e.g., movies, surveillance footage, egocentric videos, cartoons, game videos, etc., which reflects the models' LVU performances in different scenarios. 3) The development of diversified evaluation tasks, which enables a comprehensive examination of MLLMs' key abilities in long-video understanding. The empirical study with 20 latest MLLMs reveals significant room for improvement in today's technique, as all existing methods struggle with most of the evaluation tasks and exhibit severe performance degradation when handling longer videos. Additionally, it suggests that factors such as context length, image-understanding quality, and the choice of LLM backbone can play critical roles in future advancements. We anticipate that MLVU will advance the research of long video understanding by providing a comprehensive and in-depth analysis of MLLMs.
Authors: Haizhong Zheng, Elisa Tsai, Yifu Lu, Jiachen Sun, Brian R. Bartoldson, Bhavya Kailkhura, Atul Prakash
Abstract: High-quality human-annotated data is crucial for modern deep learning pipelines, yet the human annotation process is both costly and time-consuming. Given a constrained human labeling budget, selecting an informative and representative data subset for labeling can significantly reduce human annotation effort. Well-performing state-of-the-art (SOTA) coreset selection methods require ground-truth labels over the whole dataset, failing to reduce the human labeling burden. Meanwhile, SOTA label-free coreset selection methods deliver inferior performance due to poor geometry-based scores. In this paper, we introduce ELFS, a novel label-free coreset selection method. ELFS employs deep clustering to estimate data difficulty scores without ground-truth labels. Furthermore, ELFS uses a simple but effective double-end pruning method to mitigate bias on calculated scores, which further improves the performance on selected coresets. We evaluate ELFS on five vision benchmarks and show that ELFS consistently outperforms SOTA label-free baselines. For instance, at a 90% pruning rate, ELFS surpasses the best-performing baseline by 5.3% on CIFAR10 and 7.1% on CIFAR100. Moreover, ELFS even achieves comparable performance to supervised coreset selection at low pruning rates (e.g., 30% and 50%) on CIFAR10 and ImageNet-1K.
Authors: Ye Tian, Ling Yang, Haotian Yang, Yuan Gao, Yufan Deng, Jingmin Chen, Xintao Wang, Zhaochen Yu, Xin Tao, Pengfei Wan, Di Zhang, Bin Cui
Abstract: Diffusion models have demonstrated great success in text-to-video (T2V) generation. However, existing methods may face challenges when handling complex (long) video generation scenarios that involve multiple objects or dynamic changes in object numbers. To address these limitations, we propose VideoTetris, a novel framework that enables compositional T2V generation. Specifically, we propose spatio-temporal compositional diffusion to precisely follow complex textual semantics by manipulating and composing the attention maps of denoising networks spatially and temporally. Moreover, we propose an enhanced video data preprocessing to enhance the training data regarding motion dynamics and prompt understanding, equipped with a new reference frame attention mechanism to improve the consistency of auto-regressive video generation. Extensive experiments demonstrate that our VideoTetris achieves impressive qualitative and quantitative results in compositional T2V generation. Code is available at: https://github.com/YangLing0818/VideoTetris
Authors: Jackson Arnold, Sophia Rossi, Chloe Petrosino, Ethan Mitchell, Sanjeev J. Koppal
Abstract: Hyperspectral image segmentation is crucial for many fields such as agriculture, remote sensing, biomedical imaging, battlefield sensing and astronomy. However, the challenge of hyper and multi spectral imaging is its large data footprint. We propose both a novel camera design and a vision transformer-based (ViT) algorithm that alleviate both the captured data footprint and the computational load for hyperspectral segmentation. Our camera is able to adaptively sample image regions or patches at different resolutions, instead of capturing the entire hyperspectral cube at one high resolution. Our segmentation algorithm works in concert with the camera, applying ViT-based segmentation only to adaptively selected patches. We show results both in simulation and on a real hardware platform demonstrating both accurate segmentation results and reduced computational burden.
Authors: Jiayi Guo, Junhao Zhao, Chunjiang Ge, Chaoqun Du, Zanlin Ni, Shiji Song, Humphrey Shi, Gao Huang
Abstract: Test-time adaptation (TTA) aims to enhance the performance of source-domain pretrained models when tested on unknown shifted target domains. Traditional TTA methods primarily adapt model weights based on target data streams, making model performance sensitive to the amount and order of target data. Recently, diffusion-driven TTA methods have demonstrated strong performance by using an unconditional diffusion model, which is also trained on the source domain to transform target data into synthetic data as a source domain projection. This allows the source model to make predictions without weight adaptation. In this paper, we argue that the domains of the source model and the synthetic data in diffusion-driven TTA methods are not aligned. To adapt the source model to the synthetic domain of the unconditional diffusion model, we introduce a Synthetic-Domain Alignment (SDA) framework to fine-tune the source model with synthetic data. Specifically, we first employ a conditional diffusion model to generate labeled samples, creating a synthetic dataset. Subsequently, we use the aforementioned unconditional diffusion model to add noise to and denoise each sample before fine-tuning. This process mitigates the potential domain gap between the conditional and unconditional models. Extensive experiments across various models and benchmarks demonstrate that SDA achieves superior domain alignment and consistently outperforms existing diffusion-driven TTA methods. Our code is available at https://github.com/SHI-Labs/Diffusion-Driven-Test-Time-Adaptation-via-Synthetic-Domain-Alignment.
URLs: https://github.com/SHI-Labs/Diffusion-Driven-Test-Time-Adaptation-via-Synthetic-Domain-Alignment.
Authors: Kaichen Zhou
Abstract: This paper addresses the challenge of reconstructing surfaces from sparse view inputs, where ambiguity and occlusions due to missing information pose significant hurdles. We present a novel approach, named EpiS, that incorporates Epipolar information into the reconstruction process. Existing methods in sparse-view neural surface learning have mainly focused on mean and variance considerations using cost volumes for feature extraction. In contrast, our method aggregates coarse information from the cost volume into Epipolar features extracted from multiple source views, enabling the generation of fine-grained Signal Distance Function (SDF)-aware features. Additionally, we employ an attention mechanism along the line dimension to facilitate feature fusion based on the SDF feature. Furthermore, to address the information gaps in sparse conditions, we integrate depth information from monocular depth estimation using global and local regularization techniques. The global regularization utilizes a triplet loss function, while the local regularization employs a derivative loss function. Extensive experiments demonstrate that our approach outperforms state-of-the-art methods, especially in cases with sparse and generalizable conditions.
Authors: Benedikt Alkin, Maximilian Beck, Korbinian P\"oppel, Sepp Hochreiter, Johannes Brandstetter
Abstract: Transformers are widely used as generic backbones in computer vision, despite initially introduced for natural language processing. Recently, the Long Short-Term Memory (LSTM) has been extended to a scalable and performant architecture - the xLSTM - which overcomes long-standing LSTM limitations via exponential gating and parallelizable matrix memory structure. In this report, we introduce Vision-LSTM (ViL), an adaption of the xLSTM building blocks to computer vision. ViL comprises a stack of xLSTM blocks where odd blocks process the sequence of patch tokens from top to bottom while even blocks go from bottom to top. Experiments show that ViL holds promise to be further deployed as new generic backbone for computer vision architectures.
Authors: Sergey Zakharov, Katherine Liu, Adrien Gaidon, Rares Ambrus
Abstract: The common trade-offs of state-of-the-art methods for multi-shape representation (a single model "packing" multiple objects) involve trading modeling accuracy against memory and storage. We show how to encode multiple shapes represented as continuous neural fields with a higher degree of precision than previously possible and with low memory usage. Key to our approach is a recursive hierarchical formulation that exploits object self-similarity, leading to a highly compressed and efficient shape latent space. Thanks to the recursive formulation, our method supports spatial and global-to-local latent feature fusion without needing to initialize and maintain auxiliary data structures, while still allowing for continuous field queries to enable applications such as raytracing. In experiments on a set of diverse datasets, we provide compelling qualitative results and demonstrate state-of-the-art multi-scene reconstruction and compression results with a single network per dataset.
Authors: Luca Eyring, Shyamgopal Karthik, Karsten Roth, Alexey Dosovitskiy, Zeynep Akata
Abstract: Text-to-Image (T2I) models have made significant advancements in recent years, but they still struggle to accurately capture intricate details specified in complex compositional prompts. While fine-tuning T2I models with reward objectives has shown promise, it suffers from "reward hacking" and may not generalize well to unseen prompt distributions. In this work, we propose Reward-based Noise Optimization (ReNO), a novel approach that enhances T2I models at inference by optimizing the initial noise based on the signal from one or multiple human preference reward models. Remarkably, solving this optimization problem with gradient ascent for 50 iterations yields impressive results on four different one-step models across two competitive benchmarks, T2I-CompBench and GenEval. Within a computational budget of 20-50 seconds, ReNO-enhanced one-step models consistently surpass the performance of all current open-source Text-to-Image models. Extensive user studies demonstrate that our model is preferred nearly twice as often compared to the popular SDXL model and is on par with the proprietary Stable Diffusion 3 with 8B parameters. Moreover, given the same computational resources, a ReNO-optimized one-step model outperforms widely-used open-source models such as SDXL and PixArt-$\alpha$, highlighting the efficiency and effectiveness of ReNO in enhancing T2I model performance at inference time. Code is available at https://github.com/ExplainableML/ReNO.
Authors: Zhanhao Liang, Yuhui Yuan, Shuyang Gu, Bohan Chen, Tiankai Hang, Ji Li, Liang Zheng
Abstract: Recently, Direct Preference Optimization (DPO) has extended its success from aligning large language models (LLMs) to aligning text-to-image diffusion models with human preferences. Unlike most existing DPO methods that assume all diffusion steps share a consistent preference order with the final generated images, we argue that this assumption neglects step-specific denoising performance and that preference labels should be tailored to each step's contribution. To address this limitation, we propose Step-aware Preference Optimization (SPO), a novel post-training approach that independently evaluates and adjusts the denoising performance at each step, using a step-aware preference model and a step-wise resampler to ensure accurate step-aware supervision. Specifically, at each denoising step, we sample a pool of images, find a suitable win-lose pair, and, most importantly, randomly select a single image from the pool to initialize the next denoising step. This step-wise resampler process ensures the next win-lose image pair comes from the same image, making the win-lose comparison independent of the previous step. To assess the preferences at each step, we train a separate step-aware preference model that can be applied to both noisy and clean images. Our experiments with Stable Diffusion v1.5 and SDXL demonstrate that SPO significantly outperforms the latest Diffusion-DPO in aligning generated images with complex, detailed prompts and enhancing aesthetics, while also achieving more than 20x times faster in training efficiency. Code and model: https://rockeycoss.github.io/spo.github.io/
Authors: Jiyao Zhang, Weiyao Huang, Bo Peng, Mingdong Wu, Fei Hu, Zijian Chen, Bo Zhao, Hao Dong
Abstract: 6D Object Pose Estimation is a crucial yet challenging task in computer vision, suffering from a significant lack of large-scale datasets. This scarcity impedes comprehensive evaluation of model performance, limiting research advancements. Furthermore, the restricted number of available instances or categories curtails its applications. To address these issues, this paper introduces Omni6DPose, a substantial dataset characterized by its diversity in object categories, large scale, and variety in object materials. Omni6DPose is divided into three main components: ROPE (Real 6D Object Pose Estimation Dataset), which includes 332K images annotated with over 1.5M annotations across 581 instances in 149 categories; SOPE(Simulated 6D Object Pose Estimation Dataset), consisting of 475K images created in a mixed reality setting with depth simulation, annotated with over 5M annotations across 4162 instances in the same 149 categories; and the manually aligned real scanned objects used in both ROPE and SOPE. Omni6DPose is inherently challenging due to the substantial variations and ambiguities. To address this challenge, we introduce GenPose++, an enhanced version of the SOTA category-level pose estimation framework, incorporating two pivotal improvements: Semantic-aware feature extraction and Clustering-based aggregation. Moreover, we provide a comprehensive benchmarking analysis to evaluate the performance of previous methods on this large-scale dataset in the realms of 6D object pose estimation and pose tracking.
Authors: Zeyue Tian, Zhaoyang Liu, Ruibin Yuan, Jiahao Pan, Xiaoqiang Huang, Qifeng Liu, Xu Tan, Qifeng Chen, Wei Xue, Yike Guo
Abstract: In this work, we systematically study music generation conditioned solely on the video. First, we present a large-scale dataset comprising 190K video-music pairs, including various genres such as movie trailers, advertisements, and documentaries. Furthermore, we propose VidMuse, a simple framework for generating music aligned with video inputs. VidMuse stands out by producing high-fidelity music that is both acoustically and semantically aligned with the video. By incorporating local and global visual cues, VidMuse enables the creation of musically coherent audio tracks that consistently match the video content through Long-Short-Term modeling. Through extensive experiments, VidMuse outperforms existing models in terms of audio quality, diversity, and audio-visual alignment. The code and datasets will be available at https://github.com/ZeyueT/VidMuse/.
Authors: Qihao Liu, Yi Zhang, Song Bai, Adam Kortylewski, Alan Yuille
Abstract: We present DIRECT-3D, a diffusion-based 3D generative model for creating high-quality 3D assets (represented by Neural Radiance Fields) from text prompts. Unlike recent 3D generative models that rely on clean and well-aligned 3D data, limiting them to single or few-class generation, our model is directly trained on extensive noisy and unaligned `in-the-wild' 3D assets, mitigating the key challenge (i.e., data scarcity) in large-scale 3D generation. In particular, DIRECT-3D is a tri-plane diffusion model that integrates two innovations: 1) A novel learning framework where noisy data are filtered and aligned automatically during the training process. Specifically, after an initial warm-up phase using a small set of clean data, an iterative optimization is introduced in the diffusion process to explicitly estimate the 3D pose of objects and select beneficial data based on conditional density. 2) An efficient 3D representation that is achieved by disentangling object geometry and color features with two separate conditional diffusion models that are optimized hierarchically. Given a prompt input, our model generates high-quality, high-resolution, realistic, and complex 3D objects with accurate geometric details in seconds. We achieve state-of-the-art performance in both single-class generation and text-to-3D generation. We also demonstrate that DIRECT-3D can serve as a useful 3D geometric prior of objects, for example to alleviate the well-known Janus problem in 2D-lifting methods such as DreamFusion. The code and models are available for research purposes at: https://github.com/qihao067/direct3d.
Authors: Zhixing Zhang, Yanyu Li, Yushu Wu, Yanwu Xu, Anil Kag, Ivan Skorokhodov, Willi Menapace, Aliaksandr Siarohin, Junli Cao, Dimitris Metaxas, Sergey Tulyakov, Jian Ren
Abstract: Diffusion-based video generation models have demonstrated remarkable success in obtaining high-fidelity videos through the iterative denoising process. However, these models require multiple denoising steps during sampling, resulting in high computational costs. In this work, we propose a novel approach to obtain single-step video generation models by leveraging adversarial training to fine-tune pre-trained video diffusion models. We show that, through the adversarial training, the multi-steps video diffusion model, i.e., Stable Video Diffusion (SVD), can be trained to perform single forward pass to synthesize high-quality videos, capturing both temporal and spatial dependencies in the video data. Extensive experiments demonstrate that our method achieves competitive generation quality of synthesized videos with significantly reduced computational overhead for the denoising process (i.e., around $23\times$ speedup compared with SVD and $6\times$ speedup compared with existing works, with even better generation quality), paving the way for real-time video synthesis and editing. More visualization results are made publicly available at https://snap-research.github.io/SF-V.
Authors: Lin Chen, Xilin Wei, Jinsong Li, Xiaoyi Dong, Pan Zhang, Yuhang Zang, Zehui Chen, Haodong Duan, Bin Lin, Zhenyu Tang, Li Yuan, Yu Qiao, Dahua Lin, Feng Zhao, Jiaqi Wang
Abstract: We present the ShareGPT4Video series, aiming to facilitate the video understanding of large video-language models (LVLMs) and the video generation of text-to-video models (T2VMs) via dense and precise captions. The series comprises: 1) ShareGPT4Video, 40K GPT4V annotated dense captions of videos with various lengths and sources, developed through carefully designed data filtering and annotating strategy. 2) ShareCaptioner-Video, an efficient and capable captioning model for arbitrary videos, with 4.8M high-quality aesthetic videos annotated by it. 3) ShareGPT4Video-8B, a simple yet superb LVLM that reached SOTA performance on three advancing video benchmarks. To achieve this, taking aside the non-scalable costly human annotators, we find using GPT4V to caption video with a naive multi-frame or frame-concatenation input strategy leads to less detailed and sometimes temporal-confused results. We argue the challenge of designing a high-quality video captioning strategy lies in three aspects: 1) Inter-frame precise temporal change understanding. 2) Intra-frame detailed content description. 3) Frame-number scalability for arbitrary-length videos. To this end, we meticulously designed a differential video captioning strategy, which is stable, scalable, and efficient for generating captions for videos with arbitrary resolution, aspect ratios, and length. Based on it, we construct ShareGPT4Video, which contains 40K high-quality videos spanning a wide range of categories, and the resulting captions encompass rich world knowledge, object attributes, camera movements, and crucially, detailed and precise temporal descriptions of events. Based on ShareGPT4Video, we further develop ShareCaptioner-Video, a superior captioner capable of efficiently generating high-quality captions for arbitrary videos...
Authors: Xizhou Zhu, Xue Yang, Zhaokai Wang, Hao Li, Wenhan Dou, Junqi Ge, Lewei Lu, Yu Qiao, Jifeng Dai
Abstract: Image pyramids are commonly used in modern computer vision tasks to obtain multi-scale features for precise understanding of images. However, image pyramids process multiple resolutions of images using the same large-scale model, which requires significant computational cost. To overcome this issue, we propose a novel network architecture known as the Parameter-Inverted Image Pyramid Networks (PIIP). Our core idea is to use models with different parameter sizes to process different resolution levels of the image pyramid, thereby balancing computational efficiency and performance. Specifically, the input to PIIP is a set of multi-scale images, where higher resolution images are processed by smaller networks. We further propose a feature interaction mechanism to allow features of different resolutions to complement each other and effectively integrate information from different spatial scales. Extensive experiments demonstrate that the PIIP achieves superior performance in tasks such as object detection, segmentation, and image classification, compared to traditional image pyramid methods and single-branch networks, while reducing computational cost. Notably, when applying our method on a large-scale vision foundation model InternViT-6B, we improve its performance by 1%-2% on detection and segmentation with only 40%-60% of the original computation. These results validate the effectiveness of the PIIP approach and provide a new technical direction for future vision computing tasks. Our code and models are available at https://github.com/OpenGVLab/PIIP.
Authors: Sebastian Loeschcke, Dan Wang, Christian Leth-Espensen, Serge Belongie, Michael J. Kastoryano, Sagie Benaim
Abstract: The ability to learn compact, high-quality, and easy-to-optimize representations for visual data is paramount to many applications such as novel view synthesis and 3D reconstruction. Recent work has shown substantial success in using tensor networks to design such compact and high-quality representations. However, the ability to optimize tensor-based representations, and in particular, the highly compact tensor train representation, is still lacking. This has prevented practitioners from deploying the full potential of tensor networks for visual data. To this end, we propose 'Prolongation Upsampling Tensor Train (PuTT)', a novel method for learning tensor train representations in a coarse-to-fine manner. Our method involves the prolonging or `upsampling' of a learned tensor train representation, creating a sequence of 'coarse-to-fine' tensor trains that are incrementally refined. We evaluate our representation along three axes: (1). compression, (2). denoising capability, and (3). image completion capability. To assess these axes, we consider the tasks of image fitting, 3D fitting, and novel view synthesis, where our method shows an improved performance compared to state-of-the-art tensor-based methods. For full results see our project webpage: https://sebulo.github.io/PuTT_website/
Authors: Yang Sui, Yanyu Li, Anil Kag, Yerlan Idelbayev, Junli Cao, Ju Hu, Dhritiman Sagar, Bo Yuan, Sergey Tulyakov, Jian Ren
Abstract: Diffusion-based image generation models have achieved great success in recent years by showing the capability of synthesizing high-quality content. However, these models contain a huge number of parameters, resulting in a significantly large model size. Saving and transferring them is a major bottleneck for various applications, especially those running on resource-constrained devices. In this work, we develop a novel weight quantization method that quantizes the UNet from Stable Diffusion v1.5 to 1.99 bits, achieving a model with 7.9X smaller size while exhibiting even better generation quality than the original one. Our approach includes several novel techniques, such as assigning optimal bits to each layer, initializing the quantized model for better performance, and improving the training strategy to dramatically reduce quantization error. Furthermore, we extensively evaluate our quantized model across various benchmark datasets and through human evaluation to demonstrate its superior generation quality.
Authors: Lingchen Meng, Jianwei Yang, Rui Tian, Xiyang Dai, Zuxuan Wu, Jianfeng Gao, Yu-Gang Jiang
Abstract: Most large multimodal models (LMMs) are implemented by feeding visual tokens as a sequence into the first layer of a large language model (LLM). The resulting architecture is simple but significantly increases computation and memory costs, as it has to handle a large number of additional tokens in its input layer. This paper presents a new architecture DeepStack for LMMs. Considering $N$ layers in the language and vision transformer of LMMs, we stack the visual tokens into $N$ groups and feed each group to its aligned transformer layer \textit{from bottom to top}. Surprisingly, this simple method greatly enhances the power of LMMs to model interactions among visual tokens across layers but with minimal additional cost. We apply DeepStack to both language and vision transformer in LMMs, and validate the effectiveness of DeepStack LMMs with extensive empirical results. Using the same context length, our DeepStack 7B and 13B parameters surpass their counterparts by \textbf{2.7} and \textbf{2.9} on average across \textbf{9} benchmarks, respectively. Using only one-fifth of the context length, DeepStack rivals closely to the counterparts that use the full context length. These gains are particularly pronounced on high-resolution tasks, e.g., \textbf{4.2}, \textbf{11.0}, and \textbf{4.0} improvements on TextVQA, DocVQA, and InfoVQA compared to LLaVA-1.5-7B, respectively. We further apply DeepStack to vision transformer layers, which brings us a similar amount of improvements, \textbf{3.8} on average compared with LLaVA-1.5-7B.
Authors: Quynh Phung, Songwei Ge, Jia-Bin Huang
Abstract: Despite the advances in text-to-image synthesis, particularly with diffusion models, generating visual instructions that require consistent representation and smooth state transitions of objects across sequential steps remains a formidable challenge. This paper introduces a simple, training-free framework to tackle the issues, capitalizing on the advancements in diffusion models and large language models (LLMs). Our approach systematically integrates text comprehension and image generation to ensure visual instructions are visually appealing and maintain consistency and accuracy throughout the instruction sequence. We validate the effectiveness by testing multi-step instructions and comparing the text alignment and consistency with several baselines. Our experiments show that our approach can visualize coherent and visually pleasing instructions
Authors: Fangfu Liu, Hanyang Wang, Shunyu Yao, Shengjun Zhang, Jie Zhou, Yueqi Duan
Abstract: In recent years, there has been rapid development in 3D generation models, opening up new possibilities for applications such as simulating the dynamic movements of 3D objects and customizing their behaviors. However, current 3D generative models tend to focus only on surface features such as color and shape, neglecting the inherent physical properties that govern the behavior of objects in the real world. To accurately simulate physics-aligned dynamics, it is essential to predict the physical properties of materials and incorporate them into the behavior prediction process. Nonetheless, predicting the diverse materials of real-world objects is still challenging due to the complex nature of their physical attributes. In this paper, we propose \textbf{Physics3D}, a novel method for learning various physical properties of 3D objects through a video diffusion model. Our approach involves designing a highly generalizable physical simulation system based on a viscoelastic material model, which enables us to simulate a wide range of materials with high-fidelity capabilities. Moreover, we distill the physical priors from a video diffusion model that contains more understanding of realistic object materials. Extensive experiments demonstrate the effectiveness of our method with both elastic and plastic materials. Physics3D shows great potential for bridging the gap between the physical world and virtual neural space, providing a better integration and application of realistic physical principles in virtual environments. Project page: https://liuff19.github.io/Physics3D.
Authors: Jiaming Liu, Mengzhen Liu, Zhenyu Wang, Lily Lee, Kaichen Zhou, Pengju An, Senqiao Yang, Renrui Zhang, Yandong Guo, Shanghang Zhang
Abstract: A fundamental objective in robot manipulation is to enable models to comprehend visual scenes and execute actions. Although existing robot Multimodal Large Language Models (MLLMs) can handle a range of basic tasks, they still face challenges in two areas: 1) inadequate reasoning ability to tackle complex tasks, and 2) high computational costs for MLLM fine-tuning and inference. The recently proposed state space model (SSM) known as Mamba demonstrates promising capabilities in non-trivial sequence modeling with linear inference complexity. Inspired by this, we introduce RoboMamba, an end-to-end robotic MLLM that leverages the Mamba model to deliver both robotic reasoning and action capabilities, while maintaining efficient fine-tuning and inference. Specifically, we first integrate the vision encoder with Mamba, aligning visual data with language embedding through co-training, empowering our model with visual common sense and robot-related reasoning. To further equip RoboMamba with action pose prediction abilities, we explore an efficient fine-tuning strategy with a simple policy head. We find that once RoboMamba possesses sufficient reasoning capability, it can acquire manipulation skills with minimal fine-tuning parameters (0.1\% of the model) and time (20 minutes). In experiments, RoboMamba demonstrates outstanding reasoning capabilities on general and robotic evaluation benchmarks. Meanwhile, our model showcases impressive pose prediction results in both simulation and real-world experiments, achieving inference speeds 7 times faster than existing robot MLLMs. Our project web page: https://sites.google.com/view/robomamba-web
Authors: Fangjinhua Wang, Xudong Jiang, Silvano Galliani, Christoph Vogel, Marc Pollefeys
Abstract: Scene coordinate regression (SCR) methods are a family of visual localization methods that directly regress 2D-3D matches for camera pose estimation. They are effective in small-scale scenes but face significant challenges in large-scale scenes that are further amplified in the absence of ground truth 3D point clouds for supervision. Here, the model can only rely on reprojection constraints and needs to implicitly triangulate the points. The challenges stem from a fundamental dilemma: The network has to be invariant to observations of the same landmark at different viewpoints and lighting conditions, etc., but at the same time discriminate unrelated but similar observations. The latter becomes more relevant and severe in larger scenes. In this work, we tackle this problem by introducing the concept of co-visibility to the network. We propose GLACE, which integrates pre-trained global and local encodings and enables SCR to scale to large scenes with only a single small-sized network. Specifically, we propose a novel feature diffusion technique that implicitly groups the reprojection constraints with co-visibility and avoids overfitting to trivial solutions. Additionally, our position decoder parameterizes the output positions for large-scale scenes more effectively. Without using 3D models or depth maps for supervision, our method achieves state-of-the-art results on large-scale scenes with a low-map-size model. On Cambridge landmarks, with a single model, we achieve 17% lower median position error than Poker, the ensemble variant of the state-of-the-art SCR method ACE. Code is available at: https://github.com/cvg/glace.
Authors: Yossi Gandelsman, Alexei A. Efros, Jacob Steinhardt
Abstract: We interpret the function of individual neurons in CLIP by automatically describing them using text. Analyzing the direct effects (i.e. the flow from a neuron through the residual stream to the output) or the indirect effects (overall contribution) fails to capture the neurons' function in CLIP. Therefore, we present the "second-order lens", analyzing the effect flowing from a neuron through the later attention heads, directly to the output. We find that these effects are highly selective: for each neuron, the effect is significant for <2% of the images. Moreover, each effect can be approximated by a single direction in the text-image space of CLIP. We describe neurons by decomposing these directions into sparse sets of text representations. The sets reveal polysemantic behavior - each neuron corresponds to multiple, often unrelated, concepts (e.g. ships and cars). Exploiting this neuron polysemy, we mass-produce "semantic" adversarial examples by generating images with concepts spuriously correlated to the incorrect class. Additionally, we use the second-order effects for zero-shot segmentation and attribute discovery in images. Our results indicate that a scalable understanding of neurons can be used for model deception and for introducing new model capabilities.
Authors: Chenxin Tao, Xizhou Zhu, Shiqian Su, Lewei Lu, Changyao Tian, Xuan Luo, Gao Huang, Hongsheng Li, Yu Qiao, Jie Zhou, Jifeng Dai
Abstract: Modality differences have led to the development of heterogeneous architectures for vision and language models. While images typically require 2D non-causal modeling, texts utilize 1D causal modeling. This distinction poses significant challenges in constructing unified multi-modal models. This paper explores the feasibility of representing images using 1D causal modeling. We identify an "over-focus" issue in existing 1D causal vision models, where attention overly concentrates on a small proportion of visual tokens. The issue of "over-focus" hinders the model's ability to extract diverse visual features and to receive effective gradients for optimization. To address this, we propose De-focus Attention Networks, which employ learnable bandpass filters to create varied attention patterns. During training, large and scheduled drop path rates, and an auxiliary loss on globally pooled features for global understanding tasks are introduced. These two strategies encourage the model to attend to a broader range of tokens and enhance network optimization. Extensive experiments validate the efficacy of our approach, demonstrating that 1D causal visual representation can perform comparably to 2D non-causal representation in tasks such as global perception, dense prediction, and multi-modal understanding. Code is released at https://github.com/OpenGVLab/De-focus-Attention-Networks.
URLs: https://github.com/OpenGVLab/De-focus-Attention-Networks.
Authors: Stanislaw Szymanowicz, Eldar Insafutdinov, Chuanxia Zheng, Dylan Campbell, Jo\~ao F. Henriques, Christian Rupprecht, Andrea Vedaldi
Abstract: In this paper, we propose Flash3D, a method for scene reconstruction and novel view synthesis from a single image which is both very generalisable and efficient. For generalisability, we start from a "foundation" model for monocular depth estimation and extend it to a full 3D shape and appearance reconstructor. For efficiency, we base this extension on feed-forward Gaussian Splatting. Specifically, we predict a first layer of 3D Gaussians at the predicted depth, and then add additional layers of Gaussians that are offset in space, allowing the model to complete the reconstruction behind occlusions and truncations. Flash3D is very efficient, trainable on a single GPU in a day, and thus accessible to most researchers. It achieves state-of-the-art results when trained and tested on RealEstate10k. When transferred to unseen datasets like NYU it outperforms competitors by a large margin. More impressively, when transferred to KITTI, Flash3D achieves better PSNR than methods trained specifically on that dataset. In some instances, it even outperforms recent methods that use multiple views as input. Code, models, demo, and more results are available at https://www.robots.ox.ac.uk/~vgg/research/flash3d/.
Authors: Luca Bartolomei, Matteo Poggi, Fabio Tosi, Andrea Conti, Stefano Mattoccia
Abstract: This paper presents a novel general-purpose stereo and depth data fusion paradigm that mimics the active stereo principle by replacing the unreliable physical pattern projector with a depth sensor. It works by projecting virtual patterns consistent with the scene geometry onto the left and right images acquired by a conventional stereo camera, using the sparse hints obtained from a depth sensor, to facilitate the visual correspondence. Purposely, any depth sensing device can be seamlessly plugged into our framework, enabling the deployment of a virtual active stereo setup in any possible environment and overcoming the severe limitations of physical pattern projection, such as the limited working range and environmental conditions. Exhaustive experiments on indoor and outdoor datasets featuring both long and close range, including those providing raw, unfiltered depth hints from off-the-shelf depth sensors, highlight the effectiveness of our approach in notably boosting the robustness and accuracy of algorithms and deep stereo without any code modification and even without re-training. Additionally, we assess the performance of our strategy on active stereo evaluation datasets with conventional pattern projection. Indeed, in all these scenarios, our virtual pattern projection paradigm achieves state-of-the-art performance. The source code is available at: https://github.com/bartn8/vppstereo.
Authors: Pietro Mazzaglia, Taco Cohen, Daniel Dijkman
Abstract: Robotic affordances, providing information about what actions can be taken in a given situation, can aid robotic manipulation. However, learning about affordances requires expensive large annotated datasets of interactions or demonstrations. In this work, we argue that well-directed interactions with the environment can mitigate this problem and propose an information-based measure to augment the agent's objective and accelerate the affordance discovery process. We provide a theoretical justification of our approach and we empirically validate the approach both in simulation and real-world tasks. Our method, which we dub IDA, enables the efficient discovery of visual affordances for several action primitives, such as grasping, stacking objects, or opening drawers, strongly improving data efficiency in simulation, and it allows us to learn grasping affordances in a small number of interactions, on a real-world setup with a UFACTORY XArm 6 robot arm.
Authors: Cyprien Qu\'em\'eneur, Soumaya Cherkaoui
Abstract: The Internet of Vehicles (IoV) emerges as a pivotal component for autonomous driving and intelligent transportation systems (ITS), by enabling low-latency big data processing in a dense interconnected network that comprises vehicles, infrastructures, pedestrians and the cloud. Autonomous vehicles are heavily reliant on machine learning (ML) and can strongly benefit from the wealth of sensory data generated at the edge, which calls for measures to reconcile model training with preserving the privacy of sensitive user data. Federated learning (FL) stands out as a promising solution to train sophisticated ML models in vehicular networks while protecting the privacy of road users and mitigating communication overhead. This paper examines the federated optimization of the cutting-edge YOLOv7 model to tackle real-time object detection amid data heterogeneity, encompassing unbalancedness, concept drift, and label distribution skews. To this end, we introduce FedPylot, a lightweight MPI-based prototype to simulate federated object detection experiments on high-performance computing (HPC) systems, where we safeguard server-client communications using hybrid encryption. Our study factors in accuracy, communication cost, and inference speed, thereby presenting a balanced approach to the challenges faced by autonomous vehicles. We demonstrate promising results for the applicability of FL in IoV and hope that FedPylot will provide a basis for future research into federated real-time object detection. The source code is available at https://github.com/cyprienquemeneur/fedpylot.
Authors: Benjamin Hou, Qingqing Zhu, Tejas Sudarshan Mathai, Qiao Jin, Zhiyong Lu, Ronald M. Summers
Abstract: In this paper, we introduce DRR-RATE, a large-scale synthetic chest X-ray dataset derived from the recently released CT-RATE dataset. DRR-RATE comprises of 50,188 frontal Digitally Reconstructed Radiographs (DRRs) from 21,304 unique patients. Each image is paired with a corresponding radiology text report and binary labels for 18 pathology classes. Given the controllable nature of DRR generation, it facilitates the inclusion of lateral view images and images from any desired viewing position. This opens up avenues for research into new and novel multimodal applications involving paired CT, X-ray images from various views, text, and binary labels. We demonstrate the applicability of DRR-RATE alongside existing large-scale chest X-ray resources, notably the CheXpert dataset and CheXnet model. Experiments demonstrate that CheXnet, when trained and tested on the DRR-RATE dataset, achieves sufficient to high AUC scores for the six common pathologies cited in common literature: Atelectasis, Cardiomegaly, Consolidation, Lung Lesion, Lung Opacity, and Pleural Effusion. Additionally, CheXnet trained on the CheXpert dataset can accurately identify several pathologies, even when operating out of distribution. This confirms that the generated DRR images effectively capture the essential pathology features from CT images. The dataset and labels are publicly accessible at https://huggingface.co/datasets/farrell236/DRR-RATE.
Authors: Faisal Tareque Shohan, Mir Tafseer Nayeem, Samsul Islam, Abu Ubaida Akash, Shafiq Joty
Abstract: Millions of news articles published online daily can overwhelm readers. Headlines and entity (topic) tags are essential for guiding readers to decide if the content is worth their time. While headline generation has been extensively studied, tag generation remains largely unexplored, yet it offers readers better access to topics of interest. The need for conciseness in capturing readers' attention necessitates improved content selection strategies for identifying salient and relevant segments within lengthy articles, thereby guiding language models effectively. To address this, we propose to leverage auxiliary information such as images and captions embedded in the articles to retrieve relevant sentences and utilize instruction tuning with variations to generate both headlines and tags for news articles in a multilingual context. To make use of the auxiliary information, we have compiled a dataset named XL-HeadTags, which includes 20 languages across 6 diverse language families. Through extensive evaluation, we demonstrate the effectiveness of our plug-and-play multimodal-multilingual retrievers for both tasks. Additionally, we have developed a suite of tools for processing and evaluating multilingual texts, significantly contributing to the research community by enabling more accurate and efficient analysis across languages.
Authors: Yue Xu, Zhilin Lin, Yusong Qiu, Cewu Lu, Yong-Lu Li
Abstract: Though dataset distillation has witnessed rapid development in recent years, the distillation of multimodal data, e.g., image-text pairs, poses unique and under-explored challenges. Unlike unimodal data, image-text contrastive learning (ITC) data lack inherent categorization and should instead place greater emphasis on modality correspondence. In this work, we propose Low-Rank Similarity Mining (LoRS) for multimodal dataset distillation, that concurrently distills a ground truth similarity matrix with image-text pairs, and leverages low-rank factorization for efficiency and scalability. The proposed approach brings significant improvement to the existing algorithms, marking a significant contribution to the field of visual-language dataset distillation. We advocate adopting LoRS as a foundational synthetic data setup for image-text dataset distillation. Our code is available at https://github.com/silicx/LoRS_Distill.
Authors: Stefan Gerd Fritsch, Cennet Oguz, Vitor Fortes Rey, Lala Ray, Maximilian Kiefer-Emmanouilidis, Paul Lukowicz
Abstract: Human Activity Recognition is a longstanding problem in AI with applications in a broad range of areas: from healthcare, sports and fitness, security, and human computer interaction to robotics. The performance of HAR in real-world settings is strongly dependent on the type and quality of the input signal that can be acquired. Given an unobstructed, high-quality camera view of a scene, computer vision systems, in particular in conjunction with foundational models (e.g., CLIP), can today fairly reliably distinguish complex activities. On the other hand, recognition using modalities such as wearable sensors (which are often more broadly available, e.g, in mobile phones and smartwatches) is a more difficult problem, as the signals often contain less information and labeled training data is more difficult to acquire. In this work, we show how we can improve HAR performance across different modalities using multimodal contrastive pretraining. Our approach MuJo (Multimodal Joint Feature Space Learning), learns a multimodal joint feature space with video, language, pose, and IMU sensor data. The proposed approach combines contrastive and multitask learning methods and analyzes different multitasking strategies for learning a compact shared representation. A large dataset with parallel video, language, pose, and sensor data points is also introduced to support the research, along with an analysis of the robustness of the multimodal joint space for modal-incomplete and low-resource data. On the MM-Fit dataset, our model achieves an impressive Macro F1-Score of up to 0.992 with only 2% of the train data and 0.999 when using all available training data for classification tasks. Moreover, in the scenario where the MM-Fit dataset is unseen, we demonstrate a generalization performance of up to 0.638.
Authors: Jiaming Zhao, Wenbo Qiao, Peng Zhang, Hui Gao
Abstract: Implicit neural representations have emerged as a powerful paradigm to represent signals such as images and sounds. This approach aims to utilize neural networks to parameterize the implicit function of the signal. However, when representing implicit functions, traditional neural networks such as ReLU-based multilayer perceptrons face challenges in accurately modeling high-frequency components of signals. Recent research has begun to explore the use of Fourier Neural Networks (FNNs) to overcome this limitation. In this paper, we propose Quantum Implicit Representation Network (QIREN), a novel quantum generalization of FNNs. Furthermore, through theoretical analysis, we demonstrate that QIREN possesses a quantum advantage over classical FNNs. Lastly, we conducted experiments in signal representation, image superresolution, and image generation tasks to show the superior performance of QIREN compared to state-of-the-art (SOTA) models. Our work not only incorporates quantum advantages into implicit neural representations but also uncovers a promising application direction for Quantum Neural Networks.
Authors: Xiaosong Jia, Zhenjie Yang, Qifeng Li, Zhiyuan Zhang, Junchi Yan
Abstract: In an era marked by the rapid scaling of foundation models, autonomous driving technologies are approaching a transformative threshold where end-to-end autonomous driving (E2E-AD) emerges due to its potential of scaling up in the data-driven manner. However, existing E2E-AD methods are mostly evaluated under the open-loop log-replay manner with L2 errors and collision rate as metrics (e.g., in nuScenes), which could not fully reflect the driving performance of algorithms as recently acknowledged in the community. For those E2E-AD methods evaluated under the closed-loop protocol, they are tested in fixed routes (e.g., Town05Long and Longest6 in CARLA) with the driving score as metrics, which is known for high variance due to the unsmoothed metric function and large randomness in the long route. Besides, these methods usually collect their own data for training, which makes algorithm-level fair comparison infeasible. To fulfill the paramount need of comprehensive, realistic, and fair testing environments for Full Self-Driving (FSD), we present Bench2Drive, the first benchmark for evaluating E2E-AD systems' multiple abilities in a closed-loop manner. Bench2Drive's official training data consists of 2 million fully annotated frames, collected from 10000 short clips uniformly distributed under 44 interactive scenarios (cut-in, overtaking, detour, etc), 23 weathers (sunny, foggy, rainy, etc), and 12 towns (urban, village, university, etc) in CARLA v2. Its evaluation protocol requires E2E-AD models to pass 44 interactive scenarios under different locations and weathers which sums up to 220 routes and thus provides a comprehensive and disentangled assessment about their driving capability under different situations. We implement state-of-the-art E2E-AD models and evaluate them in Bench2Drive, providing insights regarding current status and future directions.
Authors: Minghao Yang, Linlin Gao, Pengyuan Li, Wenbo Li, Yihong Dong, Zhiying Cui
Abstract: Current structured pruning methods often result in considerable accuracy drops due to abrupt network changes and loss of information from pruned structures. To address these issues, we introduce the Decay Pruning Method (DPM), a novel smooth pruning approach with a self-rectifying mechanism. DPM consists of two key components: (i) Smooth Pruning: It converts conventional single-step pruning into multi-step smooth pruning, gradually reducing redundant structures to zero over N steps with ongoing optimization. (ii) Self-Rectifying: This procedure further enhances the aforementioned process by rectifying sub-optimal pruning based on gradient information. Our approach demonstrates strong generalizability and can be easily integrated with various existing pruning methods. We validate the effectiveness of DPM by integrating it with three popular pruning methods: OTOv2, Depgraph, and Gate Decorator. Experimental results show consistent improvements in performance compared to the original pruning methods, along with further reductions of FLOPs in most scenarios.
Authors: Adrian Galdran
Abstract: This paper describes a solution for the MedAI competition, in which participants were required to segment both polyps and surgical instruments from endoscopic images. Our approach relies on a double encoder-decoder neural network which we have previously applied for polyp segmentation, but with a series of enhancements: a more powerful encoder architecture, an improved optimization procedure, and the post-processing of segmentations based on tempered model ensembling. Experimental results show that our method produces segmentations that show a good agreement with manual delineations provided by medical experts.
Authors: Yiqun Lin, Jiewen Yang, Hualiang Wang, Xinpeng Ding, Wei Zhao, Xiaomeng Li
Abstract: Cone beam computed tomography (CBCT) is an important imaging technology widely used in medical scenarios, such as diagnosis and preoperative planning. Using fewer projection views to reconstruct CT, also known as sparse-view reconstruction, can reduce ionizing radiation and further benefit interventional radiology. Compared with sparse-view reconstruction for traditional parallel/fan-beam CT, CBCT reconstruction is more challenging due to the increased dimensionality caused by the measurement process based on cone-shaped X-ray beams. As a 2D-to-3D reconstruction problem, although implicit neural representations have been introduced to enable efficient training, only local features are considered and different views are processed equally in previous works, resulting in spatial inconsistency and poor performance on complicated anatomies. To this end, we propose C^2RV by leveraging explicit multi-scale volumetric representations to enable cross-regional learning in the 3D space. Additionally, the scale-view cross-attention module is introduced to adaptively aggregate multi-scale and multi-view features. Extensive experiments demonstrate that our C^2RV achieves consistent and significant improvement over previous state-of-the-art methods on datasets with diverse anatomy.
Authors: Adrian Galdran, Miguel A. Gonz\'alez Ballester
Abstract: As current computing capabilities increase, modern machine learning and computer vision system tend to increase in complexity, mostly by means of larger models and advanced optimization strategies. Although often neglected, in many problems there is also much to be gained by considering potential improvements in understanding and better leveraging already-available training data, including annotations. This so-called data-centric approach can lead to substantial performance increases, sometimes beyond what can be achieved by larger models. In this paper we adopt such an approach for the task of justifiable glaucoma screening from retinal images. In particular, we focus on how to combine information from multiple annotators of different skills into a tailored label smoothing scheme that allows us to better employ a large collection of fundus images, instead of discarding samples suffering from inter-rater variability. Internal validation results indicate that our bespoke label smoothing approach surpasses the performance of a standard resnet50 model and also the same model trained with conventional label smoothing techniques, in particular for the multi-label scenario of predicting clinical reasons of glaucoma likelihood in a highly imbalanced screening context. Our code is made available at github.com/agaldran/justraigs .
Authors: Firoj Alam, Abul Hasnat, Fatema Ahmed, Md Arid Hasan, Maram Hasanain
Abstract: With the rise of digital communication, memes have become a significant medium for cultural and political expression that is often used to mislead audiences. Identification of such misleading and persuasive multimodal content has become more important among various stakeholders, including social media platforms, policymakers, and the broader society as they often cause harm to individuals, organizations, and/or society. While there has been effort to develop AI-based automatic systems for resource-rich languages (e.g., English), it is relatively little to none for medium to low resource languages. In this study, we focused on developing an Arabic memes dataset with manual annotations of propagandistic content. We annotated ~6K Arabic memes collected from various social media platforms, which is a first resource for Arabic multimodal research. We provide a comprehensive analysis aiming to develop computational tools for their detection. We will make them publicly available for the community.
Authors: Jan Hagnberger, Marimuthu Kalimuthu, Daniel Musekamp, Mathias Niepert
Abstract: Transformer models are increasingly used for solving Partial Differential Equations (PDEs). Several adaptations have been proposed, all of which suffer from the typical problems of Transformers, such as quadratic memory and time complexity. Furthermore, all prevalent architectures for PDE solving lack at least one of several desirable properties of an ideal surrogate model, such as (i) generalization to PDE parameters not seen during training, (ii) spatial and temporal zero-shot super-resolution, (iii) continuous temporal extrapolation, (iv) support for 1D, 2D, and 3D PDEs, and (v) efficient inference for longer temporal rollouts. To address these limitations, we propose Vectorized Conditional Neural Fields (VCNeFs), which represent the solution of time-dependent PDEs as neural fields. Contrary to prior methods, however, VCNeFs compute, for a set of multiple spatio-temporal query points, their solutions in parallel and model their dependencies through attention mechanisms. Moreover, VCNeF can condition the neural field on both the initial conditions and the parameters of the PDEs. An extensive set of experiments demonstrates that VCNeFs are competitive with and often outperform existing ML-based surrogate models.
Authors: Junhui Li, Jutao Li, Xingsong Hou, Huake Wang, Yutao Zhang, Yujie Dun, Wenke Sun
Abstract: Deep learning-based image compression algorithms typically focus on designing encoding and decoding networks and improving the accuracy of entropy model estimation to enhance the rate-distortion (RD) performance. However, few algorithms leverage the compression distortion prior from existing compression algorithms to improve RD performance. In this paper, we propose a latent diffusion model-based remote sensing image compression (LDM-RSIC) method, which aims to enhance the final decoding quality of RS images by utilizing the generated distortion prior from a LDM. Our approach consists of two stages. In the first stage, a self-encoder learns prior from the high-quality input image. In the second stage, the prior is generated through an LDM, conditioned on the decoded image of an existing learning-based image compression algorithm, to be used as auxiliary information for generating the texture-rich enhanced image. To better utilize the prior, a channel attention and gate-based dynamic feature attention module (DFAM) is embedded into a Transformer-based multi-scale enhancement network (MEN) for image enhancement. Extensive experiments demonstrate the proposed LDM-RSIC significantly outperforms existing state-of-the-art traditional and learning-based image compression algorithms in terms of both subjective perception and objective metrics. Additionally, we use the LDM-based scheme to improve the traditional image compression algorithm JPEG2000 and obtain 32.00% bit savings on the DOTA testing set. The code will be available at https://github.com/mlkk518/LDM-RSIC.
Authors: Kun Song, Zhiquan Tan, Bochao Zou, Huimin Ma, Weiran Huang
Abstract: In this paper, we use matrix information theory as an analytical tool to analyze the dynamics of the information interplay between data representations and classification head vectors in the supervised learning process. Specifically, inspired by the theory of Neural Collapse, we introduce matrix mutual information ratio (MIR) and matrix entropy difference ratio (HDR) to assess the interactions of data representation and class classification heads in supervised learning, and we determine the theoretical optimal values for MIR and HDR when Neural Collapse happens. Our experiments show that MIR and HDR can effectively explain many phenomena occurring in neural networks, for example, the standard supervised training dynamics, linear mode connectivity, and the performance of label smoothing and pruning. Additionally, we use MIR and HDR to gain insights into the dynamics of grokking, which is an intriguing phenomenon observed in supervised training, where the model demonstrates generalization capabilities long after it has learned to fit the training data. Furthermore, we introduce MIR and HDR as loss terms in supervised and semi-supervised learning to optimize the information interactions among samples and classification heads. The empirical results provide evidence of the method's effectiveness, demonstrating that the utilization of MIR and HDR not only aids in comprehending the dynamics throughout the training process but can also enhances the training procedure itself.
Authors: Tam Thuc Do, Parham Eftekhar, Seyed Alireza Hosseini, Gene Cheung, Philip Chou
Abstract: We build interpretable and lightweight transformer-like neural networks by unrolling iterative optimization algorithms that minimize graph smoothness priors -- the quadratic graph Laplacian regularizer (GLR) and the $\ell_1$-norm graph total variation (GTV) -- subject to an interpolation constraint. The crucial insight is that a normalized signal-dependent graph learning module amounts to a variant of the basic self-attention mechanism in conventional transformers. Unlike "black-box" transformers that require learning of large key, query and value matrices to compute scaled dot products as affinities and subsequent output embeddings, resulting in huge parameter sets, our unrolled networks employ shallow CNNs to learn low-dimensional features per node to establish pairwise Mahalanobis distances and construct sparse similarity graphs. At each layer, given a learned graph, the target interpolated signal is simply a low-pass filtered output derived from the minimization of an assumed graph smoothness prior, leading to a dramatic reduction in parameter count. Experiments for two image interpolation applications verify the restoration performance, parameter efficiency and robustness to covariate shift of our graph-based unrolled networks compared to conventional transformers.
Authors: Tim Salimans, Thomas Mensink, Jonathan Heek, Emiel Hoogeboom
Abstract: We present a new method for making diffusion models faster to sample. The method distills many-step diffusion models into few-step models by matching conditional expectations of the clean data given noisy data along the sampling trajectory. Our approach extends recently proposed one-step methods to the multi-step case, and provides a new perspective by interpreting these approaches in terms of moment matching. By using up to 8 sampling steps, we obtain distilled models that outperform not only their one-step versions but also their original many-step teacher models, obtaining new state-of-the-art results on the Imagenet dataset. We also show promising results on a large text-to-image model where we achieve fast generation of high resolution images directly in image space, without needing autoencoders or upsamplers.
Authors: Pietro Mazzaglia, Nicholas Backshall, Xiao Ma, Stephen James
Abstract: Joint space and task space control are the two dominant action modes for controlling robot arms within the robot learning literature. Actions in joint space provide precise control over the robot's pose, but tend to suffer from inefficient training; actions in task space boast data-efficient training but sacrifice the ability to perform tasks in confined spaces due to limited control over the full joint configuration. This work analyses the criteria for designing action spaces for robot manipulation and introduces ER (End-effector Redundancy), a novel action space formulation that, by addressing the redundancies present in the manipulator, aims to combine the advantages of both joint and task spaces, offering fine-grained comprehensive control with overactuated robot arms whilst achieving highly efficient robot learning. We present two implementations of ER, ERAngle (ERA) and ERJoint (ERJ), and we show that ERJ in particular demonstrates superior performance across multiple settings, especially when precise control over the robot configuration is required. We validate our results both in simulated and real robotic environments.
Authors: Tamer Ahmed Eltaras, Qutaibah Malluhi, Alessandro Savino, Stefano Di Carlo, Adnan Qayyum, Junaid Qadir
Abstract: In the effort to learn from extensive collections of distributed data, federated learning has emerged as a promising approach for preserving privacy by using a gradient-sharing mechanism instead of exchanging raw data. However, recent studies show that private training data can be leaked through many gradient attacks. While previous analytical-based attacks have successfully reconstructed input data from fully connected layers, their effectiveness diminishes when applied to convolutional layers. This paper introduces an advanced data leakage method to efficiently exploit convolutional layers' gradients. We present a surprising finding: even with non-fully invertible activation functions, such as ReLU, we can analytically reconstruct training samples from the gradients. To the best of our knowledge, this is the first analytical approach that successfully reconstructs convolutional layer inputs directly from the gradients, bypassing the need to reconstruct layers' outputs. Prior research has mainly concentrated on the weight constraints of convolution layers, overlooking the significance of gradient constraints. Our findings demonstrate that existing analytical methods used to estimate the risk of gradient attacks lack accuracy. In some layers, attacks can be launched with less than 5% of the reported constraints.
Authors: Julius Hense, Mina Jamshidi Idaji, Oliver Eberle, Thomas Schnake, Jonas Dippel, Laure Ciernik, Oliver Buchstab, Andreas Mock, Frederick Klauschen, Klaus-Robert M\"uller
Abstract: Multiple instance learning (MIL) is an effective and widely used approach for weakly supervised machine learning. In histopathology, MIL models have achieved remarkable success in tasks like tumor detection, biomarker prediction, and outcome prognostication. However, MIL explanation methods are still lagging behind, as they are limited to small bag sizes or disregard instance interactions. We revisit MIL through the lens of explainable AI (XAI) and introduce xMIL, a refined framework with more general assumptions. We demonstrate how to obtain improved MIL explanations using layer-wise relevance propagation (LRP) and conduct extensive evaluation experiments on three toy settings and four real-world histopathology datasets. Our approach consistently outperforms previous explanation attempts with particularly improved faithfulness scores on challenging biomarker prediction tasks. Finally, we showcase how xMIL explanations enable pathologists to extract insights from MIL models, representing a significant advance for knowledge discovery and model debugging in digital histopathology.
Authors: Junjie Zhou, Zheng Liu, Shitao Xiao, Bo Zhao, Yongping Xiong
Abstract: Multi-modal retrieval becomes increasingly popular in practice. However, the existing retrievers are mostly text-oriented, which lack the capability to process visual information. Despite the presence of vision-language models like CLIP, the current methods are severely limited in representing the text-only and image-only data. In this work, we present a new embedding model VISTA for universal multi-modal retrieval. Our work brings forth threefold technical contributions. Firstly, we introduce a flexible architecture which extends a powerful text encoder with the image understanding capability by introducing visual token embeddings. Secondly, we develop two data generation strategies, which bring high-quality composed image-text to facilitate the training of the embedding model. Thirdly, we introduce a multi-stage training algorithm, which first aligns the visual token embedding with the text encoder using massive weakly labeled data, and then develops multi-modal representation capability using the generated composed image-text data. In our experiments, VISTA achieves superior performances across a variety of multi-modal retrieval tasks in both zero-shot and supervised settings. Our model, data, and source code are available at https://github.com/FlagOpen/FlagEmbedding.
Authors: Andy Zou, Long Phan, Justin Wang, Derek Duenas, Maxwell Lin, Maksym Andriushchenko, Rowan Wang, Zico Kolter, Matt Fredrikson, Dan Hendrycks
Abstract: AI systems can take harmful actions and are highly vulnerable to adversarial attacks. We present an approach, inspired by recent advances in representation engineering, that "short-circuits" models as they respond with harmful outputs. Existing techniques aimed at improving alignment, such as refusal training, are often bypassed. Techniques such as adversarial training try to plug these holes by countering specific attacks. As an alternative to refusal training and adversarial training, short-circuiting directly controls the representations that are responsible for harmful outputs in the first place. Our technique can be applied to both text-only and multimodal language models to prevent the generation of harmful outputs without sacrificing utility -- even in the presence of powerful unseen attacks. Notably, while adversarial robustness in standalone image recognition remains an open challenge, short-circuiting allows the larger multimodal system to reliably withstand image "hijacks" that aim to produce harmful content. Finally, we extend our approach to AI agents, demonstrating considerable reductions in the rate of harmful actions when they are under attack. Our approach represents a significant step forward in the development of reliable safeguards to harmful behavior and adversarial attacks.
Authors: Chen-Yu Yen, Raghav Singhal, Umang Sharma, Rajesh Ranganath, Sumit Chopra, Lerrel Pinto
Abstract: Magnetic Resonance (MR) imaging, despite its proven diagnostic utility, remains an inaccessible imaging modality for disease surveillance at the population level. A major factor rendering MR inaccessible is lengthy scan times. An MR scanner collects measurements associated with the underlying anatomy in the Fourier space, also known as the k-space. Creating a high-fidelity image requires collecting large quantities of such measurements, increasing the scan time. Traditionally to accelerate an MR scan, image reconstruction from under-sampled k-space data is the method of choice. However, recent works show the feasibility of bypassing image reconstruction and directly learning to detect disease directly from a sparser learned subset of the k-space measurements. In this work, we propose Adaptive Sampling for MR (ASMR), a sampling method that learns an adaptive policy to sequentially select k-space samples to optimize for target disease detection. On 6 out of 8 pathology classification tasks spanning the Knee, Brain, and Prostate MR scans, ASMR reaches within 2% of the performance of a fully sampled classifier while using only 8% of the k-space, as well as outperforming prior state-of-the-art work in k-space sampling such as EMRT, LOUPE, and DPS.
Authors: Qianlan Yang, Yu-Xiong Wang
Abstract: Training autonomous agents with sparse rewards is a long-standing problem in online reinforcement learning (RL), due to low data efficiency. Prior work overcomes this challenge by extracting useful knowledge from offline data, often accomplished through the learning of action distribution from offline data and utilizing the learned distribution to facilitate online RL. However, since the offline data are given and fixed, the extracted knowledge is inherently limited, making it difficult to generalize to new tasks. We propose a novel approach that leverages offline data to learn a generative diffusion model, coined as Adaptive Trajectory Diffuser (ATraDiff). This model generates synthetic trajectories, serving as a form of data augmentation and consequently enhancing the performance of online RL methods. The key strength of our diffuser lies in its adaptability, allowing it to effectively handle varying trajectory lengths and mitigate distribution shifts between online and offline data. Because of its simplicity, ATraDiff seamlessly integrates with a wide spectrum of RL methods. Empirical evaluation shows that ATraDiff consistently achieves state-of-the-art performance across a variety of environments, with particularly pronounced improvements in complicated settings. Our code and demo video are available at https://atradiff.github.io .
Authors: Tim Z. Xiao, Robert Bamler, Bernhard Sch\"olkopf, Weiyang Liu
Abstract: Motivated by the large progress made by large language models (LLMs), we introduce the framework of verbalized machine learning (VML). In contrast to conventional machine learning models that are typically optimized over a continuous parameter space, VML constrains the parameter space to be human-interpretable natural language. Such a constraint leads to a new perspective of function approximation, where an LLM with a text prompt can be viewed as a function parameterized by the text prompt. Guided by this perspective, we revisit classical machine learning problems, such as regression and classification, and find that these problems can be solved by an LLM-parameterized learner and optimizer. The major advantages of VML include (1) easy encoding of inductive bias: prior knowledge about the problem and hypothesis class can be encoded in natural language and fed into the LLM-parameterized learner; (2) automatic model class selection: the optimizer can automatically select a concrete model class based on data and verbalized prior knowledge, and it can update the model class during training; and (3) interpretable learner updates: the LLM-parameterized optimizer can provide explanations for why each learner update is performed. We conduct several studies to empirically evaluate the effectiveness of VML, and hope that VML can serve as a stepping stone to stronger interpretability and trustworthiness in ML.
Authors: Hongsen Liu
Abstract: We propose a single-shot method for simultaneous 3D object segmentation and 6-DOF pose estimation in pure 3D point clouds scenes based on a consensus that \emph{one point only belongs to one object}, i.e., each point has the potential power to predict the 6-DOF pose of its corresponding object. Unlike the recently proposed methods of the similar task, which rely on 2D detectors to predict the projection of 3D corners of the 3D bounding boxes and the 6-DOF pose must be estimated by a PnP like spatial transformation method, ours is concise enough not to require additional spatial transformation between different dimensions. Due to the lack of training data for many objects, the recently proposed 2D detection methods try to generate training data by using rendering engine and achieve good results. However, rendering in 3D space along with 6-DOF is relatively difficult. Therefore, we propose an augmented reality technology to generate the training data in semi-virtual reality 3D space. The key component of our method is a multi-task CNN architecture that can simultaneously predicts the 3D object segmentation and 6-DOF pose estimation in pure 3D point clouds. For experimental evaluation, we generate expanded training data for two state-of-the-arts 3D object datasets \cite{PLCHF}\cite{TLINEMOD} by using Augmented Reality technology (AR). We evaluate our proposed method on the two datasets. The results show that our method can be well generalized into multiple scenarios and provide performance comparable to or better than the state-of-the-arts.
Authors: Ruiying Lu, Ziheng Cheng, Bo Chen, Xin Yuan
Abstract: Video snapshot compressive imaging (SCI) utilizes a 2D detector to capture sequential video frames and compress them into a single measurement. Various reconstruction methods have been developed to recover the high-speed video frames from the snapshot measurement. However, most existing reconstruction methods are incapable of efficiently capturing long-range spatial and temporal dependencies, which are critical for video processing. In this paper, we propose a flexible and robust approach based on the graph neural network (GNN) to efficiently model non-local interactions between pixels in space and time regardless of the distance. Specifically, we develop a motion-aware dynamic GNN for better video representation, i.e., represent each node as the aggregation of relative neighbors under the guidance of frame-by-frame motions, which consists of motion-aware dynamic sampling, cross-scale node sampling, global knowledge integration, and graph aggregation. Extensive results on both simulation and real data demonstrate both the effectiveness and efficiency of the proposed approach, and the visualization illustrates the intrinsic dynamic sampling operations of our proposed model for boosting the video SCI reconstruction results. The code and model will be released.
Authors: Jiachen Liu, Pan Ji, Nitin Bansal, Changjiang Cai, Qingan Yan, Xiaolei Huang, Yi Xu
Abstract: We present a novel framework named PlaneMVS for 3D plane reconstruction from multiple input views with known camera poses. Most previous learning-based plane reconstruction methods reconstruct 3D planes from single images, which highly rely on single-view regression and suffer from depth scale ambiguity. In contrast, we reconstruct 3D planes with a multi-view-stereo (MVS) pipeline that takes advantage of multi-view geometry. We decouple plane reconstruction into a semantic plane detection branch and a plane MVS branch. The semantic plane detection branch is based on a single-view plane detection framework but with differences. The plane MVS branch adopts a set of slanted plane hypotheses to replace conventional depth hypotheses to perform plane sweeping strategy and finally learns pixel-level plane parameters and its planar depth map. We present how the two branches are learned in a balanced way, and propose a soft-pooling loss to associate the outputs of the two branches and make them benefit from each other. Extensive experiments on various indoor datasets show that PlaneMVS significantly outperforms state-of-the-art (SOTA) single-view plane reconstruction methods on both plane detection and 3D geometry metrics. Our method even outperforms a set of SOTA learning-based MVS methods thanks to the learned plane priors. To the best of our knowledge, this is the first work on 3D plane reconstruction within an end-to-end MVS framework. Source code: https://github.com/oppo-us-research/PlaneMVS.
Authors: Khoa D. Doan, Jianwen Xie, Yaxuan Zhu, Yang Zhao, Ping Li
Abstract: Leveraging supervised information can lead to superior retrieval performance in the image hashing domain but the performance degrades significantly without enough labeled data. One effective solution to boost performance is to employ generative models, such as Generative Adversarial Networks (GANs), to generate synthetic data in an image hashing model. However, GAN-based methods are difficult to train, which prevents the hashing approaches from jointly training the generative models and the hash functions. This limitation results in sub-optimal retrieval performance. To overcome this limitation, we propose a novel framework, the generative cooperative hashing network, which is based on energy-based cooperative learning. This framework jointly learns a powerful generative representation of the data and a robust hash function via two components: a top-down contrastive pair generator that synthesizes contrastive images and a bottom-up multipurpose descriptor that simultaneously represents the images from multiple perspectives, including probability density, hash code, latent code, and category. The two components are jointly learned via a novel likelihood-based cooperative learning scheme. We conduct experiments on several real-world datasets and show that the proposed method outperforms the competing hashing supervised methods, achieving up to 10\% relative improvement over the current state-of-the-art supervised hashing methods, and exhibits a significantly better performance in out-of-distribution retrieval.
Authors: Yaofo Chen, Yong Guo, Daihai Liao, Fanbing Lv, Hengjie Song, James Tin-Yau Kwok, Mingkui Tan
Abstract: Neural Architecture Search (NAS) aims to automatically find effective architectures within a predefined search space. However, the search space is often extremely large. As a result, directly searching in such a large search space is non-trivial and also very time-consuming. To address the above issues, in each search step, we seek to limit the search space to a small but effective subspace to boost both the search performance and search efficiency. To this end, we propose a novel Neural Architecture Search method via Dominative Subspace Mining (DSM-NAS) that finds promising architectures in automatically mined subspaces. Specifically, we first perform a global search, i.e ., dominative subspace mining, to find a good subspace from a set of candidates. Then, we perform a local search within the mined subspace to find effective architectures. More critically, we further boost search performance by taking well-designed/ searched architectures to initialize candidate subspaces. Experimental results demonstrate that DSM-NAS not only reduces the search cost but also discovers better architectures than state-of-the-art methods in various benchmark search spaces.
Authors: Abdullah Hamdi, Faisal AlZahrani, Silvio Giancola, Bernard Ghanem
Abstract: Multi-view projection techniques have shown themselves to be highly effective in achieving top-performing results in the recognition of 3D shapes. These methods involve learning how to combine information from multiple view-points. However, the camera view-points from which these views are obtained are often fixed for all shapes. To overcome the static nature of current multi-view techniques, we propose learning these view-points. Specifically, we introduce the Multi-View Transformation Network (MVTN), which uses differentiable rendering to determine optimal view-points for 3D shape recognition. As a result, MVTN can be trained end-to-end with any multi-view network for 3D shape classification. We integrate MVTN into a novel adaptive multi-view pipeline that is capable of rendering both 3D meshes and point clouds. Our approach demonstrates state-of-the-art performance in 3D classification and shape retrieval on several benchmarks (ModelNet40, ScanObjectNN, ShapeNet Core55). Further analysis indicates that our approach exhibits improved robustness to occlusion compared to other methods. We also investigate additional aspects of MVTN, such as 2D pretraining and its use for segmentation. To support further research in this area, we have released MVTorch, a PyTorch library for 3D understanding and generation using multi-view projections.
Authors: Zhenghao Zhang, Shengfan Zhang, Zhichao Wei, Zuozhuo Dai, Siyu Zhu
Abstract: The current state-of-the-art methods for unsupervised video object segmentation (UVOS) require extensive training on video datasets with mask annotations, limiting their effectiveness in handling challenging scenarios. However, the Segment Anything Model (SAM) introduces a new prompt-driven paradigm for image segmentation, offering new possibilities. In this study, we investigate SAM's potential for UVOS through different prompt strategies. We then propose UVOSAM, a mask-free paradigm for UVOS that utilizes the STD-Net tracker. STD-Net incorporates a spatial-temporal decoupled deformable attention mechanism to establish an effective correlation between intra- and inter-frame features, remarkably enhancing the quality of box prompts in complex video scenes. Extensive experiments on the DAVIS2017-unsupervised and YoutubeVIS19\&21 datasets demonstrate the superior performance of UVOSAM without mask supervision compared to existing mask-supervised methods, as well as its ability to generalize to weakly-annotated video datasets. Code can be found at https://github.com/alibaba/UVOSAM.
Authors: Jia-Qi Yang, Chenglei Dai, Dan OU, Dongshuai Li, Ju Huang, De-Chuan Zhan, Xiaoyi Zeng, Yang Yang
Abstract: With the advancement of multimedia internet, the impact of visual characteristics on the decision of users to click or not within the online retail industry is increasingly significant. Thus, incorporating visual features is a promising direction for further performance improvements in click-through rate (CTR). However, experiments on our production system revealed that simply injecting the image embeddings trained with established pre-training methods only has marginal improvements. We believe that the main advantage of existing image feature pre-training methods lies in their effectiveness for cross-modal predictions. However, this differs significantly from the task of CTR prediction in recommendation systems. In recommendation systems, other modalities of information (such as text) can be directly used as features in downstream models. Even if the performance of cross-modal prediction tasks is excellent, it is challenging to provide significant information gain for the downstream models. We argue that a visual feature pre-training method tailored for recommendation is necessary for further improvements beyond existing modality features. To this end, we propose an effective user intention reconstruction module to mine visual features related to user interests from behavior histories, which constructs a many-to-one correspondence. We further propose a contrastive training method to learn the user intentions and prevent the collapse of embedding vectors. We conduct extensive experimental evaluations on public datasets and our production system to verify that our method can learn users' visual interests. Our method achieves $0.46\%$ improvement in offline AUC and $0.88\%$ improvement in Taobao GMV (Cross Merchandise Volume) with p-value$<$0.01.
Authors: Yinghua Gao, Yiming Li, Xueluan Gong, Zhifeng Li, Shu-Tao Xia, Qian Wang
Abstract: Deep neural networks (DNNs) are vulnerable to backdoor attacks, where the adversary manipulates a small portion of training data such that the victim model predicts normally on the benign samples but classifies the triggered samples as the target class. The backdoor attack is an emerging yet threatening training-phase threat, leading to serious risks in DNN-based applications. In this paper, we revisit the trigger patterns of existing backdoor attacks. We reveal that they are either visible or not sparse and therefore are not stealthy enough. More importantly, it is not feasible to simply combine existing methods to design an effective sparse and invisible backdoor attack. To address this problem, we formulate the trigger generation as a bi-level optimization problem with sparsity and invisibility constraints and propose an effective method to solve it. The proposed method is dubbed sparse and invisible backdoor attack (SIBA). We conduct extensive experiments on benchmark datasets under different settings, which verify the effectiveness of our attack and its resistance to existing backdoor defenses. The codes for reproducing main experiments are available at \url{https://github.com/YinghuaGao/SIBA}.
Authors: Haozhe Xie, Zhaoxi Chen, Fangzhou Hong, Ziwei Liu
Abstract: 3D city generation is a desirable yet challenging task, since humans are more sensitive to structural distortions in urban environments. Additionally, generating 3D cities is more complex than 3D natural scenes since buildings, as objects of the same class, exhibit a wider range of appearances compared to the relatively consistent appearance of objects like trees in natural scenes. To address these challenges, we propose \textbf{CityDreamer}, a compositional generative model designed specifically for unbounded 3D cities. Our key insight is that 3D city generation should be a composition of different types of neural fields: 1) various building instances, and 2) background stuff, such as roads and green lands. Specifically, we adopt the bird's eye view scene representation and employ a volumetric render for both instance-oriented and stuff-oriented neural fields. The generative hash grid and periodic positional embedding are tailored as scene parameterization to suit the distinct characteristics of building instances and background stuff. Furthermore, we contribute a suite of CityGen Datasets, including OSM and GoogleEarth, which comprises a vast amount of real-world city imagery to enhance the realism of the generated 3D cities both in their layouts and appearances. CityDreamer achieves state-of-the-art performance not only in generating realistic 3D cities but also in localized editing within the generated cities.
Authors: Ningli Xu, Rongjun Qin, Debao Huang, Fabio Remondino
Abstract: Neural Radiance Fields (NeRF) offer the potential to benefit 3D reconstruction tasks, including aerial photogrammetry. However, the scalability and accuracy of the inferred geometry are not well-documented for large-scale aerial assets,since such datasets usually result in very high memory consumption and slow convergence.. In this paper, we aim to scale the NeRF on large-scael aerial datasets and provide a thorough geometry assessment of NeRF. Specifically, we introduce a location-specific sampling technique as well as a multi-camera tiling (MCT) strategy to reduce memory consumption during image loading for RAM, representation training for GPU memory, and increase the convergence rate within tiles. MCT decomposes a large-frame image into multiple tiled images with different camera models, allowing these small-frame images to be fed into the training process as needed for specific locations without a loss of accuracy. We implement our method on a representative approach, Mip-NeRF, and compare its geometry performance with threephotgrammetric MVS pipelines on two typical aerial datasets against LiDAR reference data. Both qualitative and quantitative results suggest that the proposed NeRF approach produces better completeness and object details than traditional approaches, although as of now, it still falls short in terms of accuracy.
Authors: Elahe Vahdani, Yingli Tian
Abstract: This paper tackles the challenge of point-supervised temporal action detection, wherein only a single frame is annotated for each action instance in the training set. Most of the current methods, hindered by the sparse nature of annotated points, struggle to effectively represent the continuous structure of actions or the inherent temporal and semantic dependencies within action instances. Consequently, these methods frequently learn merely the most distinctive segments of actions, leading to the creation of incomplete action proposals. This paper proposes POTLoc, a Pseudo-label Oriented Transformer for weakly-supervised Action Localization utilizing only point-level annotation. POTLoc is designed to identify and track continuous action structures via a self-training strategy. The base model begins by generating action proposals solely with point-level supervision. These proposals undergo refinement and regression to enhance the precision of the estimated action boundaries, which subsequently results in the production of `pseudo-labels' to serve as supplementary supervisory signals. The architecture of the model integrates a transformer with a temporal feature pyramid to capture video snippet dependencies and model actions of varying duration. The pseudo-labels, providing information about the coarse locations and boundaries of actions, assist in guiding the transformer for enhanced learning of action dynamics. POTLoc outperforms the state-of-the-art point-supervised methods on THUMOS'14 and ActivityNet-v1.2 datasets.
Authors: Daoyi Gao, D\'avid Rozenberszki, Stefan Leutenegger, Angela Dai
Abstract: Perceiving 3D structures from RGB images based on CAD model primitives can enable an effective, efficient 3D object-based representation of scenes. However, current approaches rely on supervision from expensive annotations of CAD models associated with real images, and encounter challenges due to the inherent ambiguities in the task -- both in depth-scale ambiguity in monocular perception, as well as inexact matches of CAD database models to real observations. We thus propose DiffCAD, the first weakly-supervised probabilistic approach to CAD retrieval and alignment from an RGB image. We formulate this as a conditional generative task, leveraging diffusion to learn implicit probabilistic models capturing the shape, pose, and scale of CAD objects in an image. This enables multi-hypothesis generation of different plausible CAD reconstructions, requiring only a few hypotheses to characterize ambiguities in depth/scale and inexact shape matches. Our approach is trained only on synthetic data, leveraging monocular depth and mask estimates to enable robust zero-shot adaptation to various real target domains. Despite being trained solely on synthetic data, our multi-hypothesis approach can even surpass the supervised state-of-the-art on the Scan2CAD dataset by 5.9% with 8 hypotheses.
Authors: Yunfei Fan, Tianyu Zhao, Guidong Wang
Abstract: Accuracy and computational efficiency are the most important metrics to Visual Inertial Navigation System (VINS). The existing VINS algorithms with either high accuracy or low computational complexity, are difficult to provide the high precision localization in resource-constrained devices. To this end, we propose a novel filter-based VINS framework named SchurVINS, which could guarantee both high accuracy by building a complete residual model and low computational complexity with Schur complement. Technically, we first formulate the full residual model where Gradient, Hessian and observation covariance are explicitly modeled. Then Schur complement is employed to decompose the full model into ego-motion residual model and landmark residual model. Finally, Extended Kalman Filter (EKF) update is implemented in these two models with high efficiency. Experiments on EuRoC and TUM-VI datasets show that our method notably outperforms state-of-the-art (SOTA) methods in both accuracy and computational complexity. The experimental code of SchurVINS is available at https://github.com/bytedance/SchurVINS.
Authors: Qiwei Tian, Chenhao Lin, Zhengyu Zhao, Qian Li, Chao Shen
Abstract: Adversarial training has achieved substantial performance in defending image retrieval against adversarial examples. However, existing studies in deep metric learning (DML) still suffer from two major limitations: weak adversary and model collapse. In this paper, we address these two limitations by proposing Collapse-Aware TRIplet DEcoupling (CA-TRIDE). Specifically, TRIDE yields a stronger adversary by spatially decoupling the perturbation targets into the anchor and the other candidates. Furthermore, CA prevents the consequential model collapse, based on a novel metric, collapseness, which is incorporated into the optimization of perturbation. We also identify two drawbacks of the existing robustness metric in image retrieval and propose a new metric for a more reasonable robustness evaluation. Extensive experiments on three datasets demonstrate that CA-TRIDE outperforms existing defense methods in both conventional and new metrics. Codes are available at https://github.com/michaeltian108/CA-TRIDE.
Authors: Xu Yang, Yingzhe Peng, Haoxuan Ma, Shuo Xu, Chi Zhang, Yucheng Han, Hanwang Zhang
Abstract: As Archimedes famously said, ``Give me a lever long enough and a fulcrum on which to place it, and I shall move the world'', in this study, we propose to use a tiny Language Model (LM), \eg, a Transformer with 67M parameters, to lever much larger Vision-Language Models (LVLMs) with 9B parameters. Specifically, we use this tiny \textbf{Lever-LM} to configure effective in-context demonstration (ICD) sequences to improve the In-Context Learinng (ICL) performance of LVLMs. Previous studies show that diverse ICD configurations like the selection and ordering of the demonstrations heavily affect the ICL performance, highlighting the significance of configuring effective ICD sequences. Motivated by this and by re-considering the the process of configuring ICD sequence, we find this is a mirror process of human sentence composition and further assume that effective ICD configurations may contain internal statistical patterns that can be captured by Lever-LM. Then a dataset with effective ICD sequences is constructed to train Lever-LM. After training, given novel queries, new ICD sequences are configured by the trained Lever-LM to solve vision-language tasks through ICL. Experiments show that these ICD sequences can improve the ICL performance of two LVLMs compared with some strong baselines in Visual Question Answering and Image Captioning, validating that Lever-LM can really capture the statistical patterns for levering LVLMs.
Authors: Wei Li, Xue Xu, Jiachen Liu, Xinyan Xiao
Abstract: Existing text-to-image diffusion models primarily generate images from text prompts. However, the inherent conciseness of textual descriptions poses challenges in faithfully synthesizing images with intricate details, such as specific entities or scenes. This paper presents UNIMO-G, a simple multimodal conditional diffusion framework that operates on multimodal prompts with interleaved textual and visual inputs, which demonstrates a unified ability for both text-driven and subject-driven image generation. UNIMO-G comprises two core components: a Multimodal Large Language Model (MLLM) for encoding multimodal prompts, and a conditional denoising diffusion network for generating images based on the encoded multimodal input. We leverage a two-stage training strategy to effectively train the framework: firstly pre-training on large-scale text-image pairs to develop conditional image generation capabilities, and then instruction tuning with multimodal prompts to achieve unified image generation proficiency. A well-designed data processing pipeline involving language grounding and image segmentation is employed to construct multi-modal prompts. UNIMO-G excels in both text-to-image generation and zero-shot subject-driven synthesis, and is notably effective in generating high-fidelity images from complex multimodal prompts involving multiple image entities.
Authors: Lennard Bodden, Franziska Schwaiger, Duc Bach Ha, Lars Kreuzberg, Sven Behnke
Abstract: In the era of AI at the edge, self-driving cars, and climate change, the need for energy-efficient, small, embedded AI is growing. Spiking Neural Networks (SNNs) are a promising approach to address this challenge, with their event-driven information flow and sparse activations. We propose Spiking CenterNet for object detection on event data. It combines an SNN CenterNet adaptation with an efficient M2U-Net-based decoder. Our model significantly outperforms comparable previous work on Prophesee's challenging GEN1 Automotive Detection Dataset while using less than half the energy. Distilling the knowledge of a non-spiking teacher into our SNN further increases performance. To the best of our knowledge, our work is the first approach that takes advantage of knowledge distillation in the field of spiking object detection.
Authors: Miguel Fainstein, Viviana Siless, Emmanuel Iarussi
Abstract: In recent years, there has been a growing interest in training Neural Networks to approximate Unsigned Distance Fields (UDFs) for representing open surfaces in the context of 3D reconstruction. However, UDFs are non-differentiable at the zero level set which leads to significant errors in distances and gradients, generally resulting in fragmented and discontinuous surfaces. In this paper, we propose to learn a hyperbolic scaling of the unsigned distance field, which defines a new Eikonal problem with distinct boundary conditions. This allows our formulation to integrate seamlessly with state-of-the-art continuously differentiable implicit neural representation networks, largely applied in the literature to represent signed distance fields. Our approach not only addresses the challenge of open surface representation but also demonstrates significant improvement in reconstruction quality and training performance. Moreover, the unlocked field's differentiability allows the accurate computation of essential topological properties such as normal directions and curvatures, pervasive in downstream tasks such as rendering. Through extensive experiments, we validate our approach across various data sets and against competitive baselines. The results demonstrate enhanced accuracy and up to an order of magnitude increase in speed compared to previous methods.
Authors: Davide Caffagni, Federico Cocchi, Luca Barsellotti, Nicholas Moratelli, Sara Sarto, Lorenzo Baraldi, Lorenzo Baraldi, Marcella Cornia, Rita Cucchiara
Abstract: Connecting text and visual modalities plays an essential role in generative intelligence. For this reason, inspired by the success of large language models, significant research efforts are being devoted to the development of Multimodal Large Language Models (MLLMs). These models can seamlessly integrate visual and textual modalities, while providing a dialogue-based interface and instruction-following capabilities. In this paper, we provide a comprehensive review of recent visual-based MLLMs, analyzing their architectural choices, multimodal alignment strategies, and training techniques. We also conduct a detailed analysis of these models across a wide range of tasks, including visual grounding, image generation and editing, visual understanding, and domain-specific applications. Additionally, we compile and describe training datasets and evaluation benchmarks, conducting comparisons among existing models in terms of performance and computational requirements. Overall, this survey offers a comprehensive overview of the current state of the art, laying the groundwork for future MLLMs.
Authors: Yaofo Chen, Shuaicheng Niu, Yaowei Wang, Shoukai Xu, Hengjie Song, Mingkui Tan
Abstract: The conventional deep learning paradigm often involves training a deep model on a server and then deploying the model or its distilled ones to resource-limited edge devices. Usually, the models shall remain fixed once deployed (at least for some period) due to the potential high cost of model adaptation for both the server and edge sides. However, in many real-world scenarios, the test environments may change dynamically (known as distribution shifts), which often results in degraded performance. Thus, one has to adapt the edge models promptly to attain promising performance. Moreover, with the increasing data collected at the edge, this paradigm also fails to further adapt the cloud model for better performance. To address these, we encounter two primary challenges: 1) the edge model has limited computation power and may only support forward propagation; 2) the data transmission budget between cloud and edge devices is limited in latency-sensitive scenarios. In this paper, we establish a Cloud-Edge Elastic Model Adaptation (CEMA) paradigm in which the edge models only need to perform forward propagation and the edge models can be adapted online. In our CEMA, to reduce the communication burden, we devise two criteria to exclude unnecessary samples from uploading to the cloud, i.e., dynamic unreliable and low-informative sample exclusion. Based on the uploaded samples, we update and distribute the affine parameters of normalization layers by distilling from the stronger foundation model to the edge model with a sample replay strategy. Extensive experimental results on ImageNet-C and ImageNet-R verify the effectiveness of our CEMA.
Authors: Tarun Kalluri, Bodhisattwa Prasad Majumder, Manmohan Chandraker
Abstract: We introduce LaGTran, a novel framework that utilizes text supervision to guide robust transfer of discriminative knowledge from labeled source to unlabeled target data with domain gaps. While unsupervised adaptation methods have been established to address this problem, they show limitations in handling challenging domain shifts due to their exclusive operation within the pixel-space. Motivated by our observation that semantically richer text modality has more favorable transfer properties, we devise a transfer mechanism to use a source-trained text-classifier to generate predictions on the target text descriptions, and utilize these predictions as supervision for the corresponding images. Our approach driven by language guidance is surprisingly easy and simple, yet significantly outperforms all prior approaches on challenging datasets like GeoNet and DomainNet, validating its extreme effectiveness. To further extend the scope of our study beyond images, we introduce a new benchmark called Ego2Exo to study ego-exo transfer in videos and find that our language-aided approach LaGTran yields significant gains in this highly challenging and non-trivial transfer setting. Code, models, and proposed datasets are publicly available at https://tarun005.github.io/lagtran/.
Authors: Yuqin Dai, Wanlu Zhu, Ronghui Li, Zeping Ren, Xiangzheng Zhou, Xiu Li, Jun Li, Jian Yang
Abstract: Creating group choreography from music has gained attention in cultural entertainment and virtual reality, aiming to coordinate visually cohesive and diverse group movements. Despite increasing interest, recent works face challenges in achieving aesthetically appealing choreography, primarily for two key issues: multi-dancer collision and single-dancer foot slide. To address these issues, we propose a Trajectory-Controllable Diffusion (TCDiff), a novel approach that harnesses non-overlapping trajectories to facilitate coherent dance movements. Specifically, to tackle dancer collisions, we introduce a Dance-Beat Navigator capable of generating trajectories for multiple dancers based on the music, complemented by a Distance-Consistency loss to maintain appropriate spacing among trajectories within a reasonable threshold. To mitigate foot sliding, we present a Footwork Adaptor that utilizes trajectory displacement from adjacent frames to enable flexible footwork, coupled with a Relative Forward-Kinematic loss to adjust the positioning of individual dancers' root nodes and joints. Extensive experiments demonstrate that our method achieves state-of-the-art results.
Authors: Philipp Wolters, Johannes Gilg, Torben Teepe, Fabian Herzog, Anouar Laouichi, Martin Hofmann, Gerhard Rigoll
Abstract: Low-cost, vision-centric 3D perception systems for autonomous driving have made significant progress in recent years, narrowing the gap to expensive LiDAR-based methods. The primary challenge in becoming a fully reliable alternative lies in robust depth prediction capabilities, as camera-based systems struggle with long detection ranges and adverse lighting and weather conditions. In this work, we introduce HyDRa, a novel camera-radar fusion architecture for diverse 3D perception tasks. Building upon the principles of dense BEV (Bird's Eye View)-based architectures, HyDRa introduces a hybrid fusion approach to combine the strengths of complementary camera and radar features in two distinct representation spaces. Our Height Association Transformer module leverages radar features already in the perspective view to produce more robust and accurate depth predictions. In the BEV, we refine the initial sparse representation by a Radar-weighted Depth Consistency. HyDRa achieves a new state-of-the-art for camera-radar fusion of 64.2 NDS (+1.8) and 58.4 AMOTA (+1.5) on the public nuScenes dataset. Moreover, our new semantically rich and spatially accurate BEV features can be directly converted into a powerful occupancy representation, beating all previous camera-based methods on the Occ3D benchmark by an impressive 3.7 mIoU. Code and models are available at https://github.com/phi-wol/hydra.
Authors: Fangqiang Ding, Yunzhou Zhu, Xiangyu Wen, Gaowen Liu, Chris Xiaoxuan Lu
Abstract: In this work, we present ThermoHands, a new benchmark for thermal image-based egocentric 3D hand pose estimation, aimed at overcoming challenges like varying lighting conditions and obstructions (e.g., handwear). The benchmark includes a multi-view and multi-spectral dataset collected from 28 subjects performing hand-object and hand-virtual interactions under diverse scenarios, accurately annotated with 3D hand poses through an automated process. We introduce a new baseline method, TherFormer, utilizing dual transformer modules for effective egocentric 3D hand pose estimation in thermal imagery. Our experimental results highlight TherFormer's leading performance and affirm thermal imaging's effectiveness in enabling robust 3D hand pose estimation in adverse conditions.
Authors: Bu Jin, Yupeng Zheng, Pengfei Li, Weize Li, Yuhang Zheng, Sujie Hu, Xinyu Liu, Jinwei Zhu, Zhijie Yan, Haiyang Sun, Kun Zhan, Peng Jia, Xiaoxiao Long, Yilun Chen, Hao Zhao
Abstract: 3D dense captioning stands as a cornerstone in achieving a comprehensive understanding of 3D scenes through natural language. It has recently witnessed remarkable achievements, particularly in indoor settings. However, the exploration of 3D dense captioning in outdoor scenes is hindered by two major challenges: 1) the domain gap between indoor and outdoor scenes, such as dynamics and sparse visual inputs, makes it difficult to directly adapt existing indoor methods; 2) the lack of data with comprehensive box-caption pair annotations specifically tailored for outdoor scenes. To this end, we introduce the new task of outdoor 3D dense captioning. As input, we assume a LiDAR point cloud and a set of RGB images captured by the panoramic camera rig. The expected output is a set of object boxes with captions. To tackle this task, we propose the TOD3Cap network, which leverages the BEV representation to generate object box proposals and integrates Relation Q-Former with LLaMA-Adapter to generate rich captions for these objects. We also introduce the TOD3Cap dataset, the largest one to our knowledge for 3D dense captioning in outdoor scenes, which contains 2.3M descriptions of 64.3K outdoor objects from 850 scenes. Notably, our TOD3Cap network can effectively localize and caption 3D objects in outdoor scenes, which outperforms baseline methods by a significant margin (+9.6 CiDEr@0.5IoU). Code, data, and models are publicly available at https://github.com/jxbbb/TOD3Cap.
Authors: Runqi Wang, Caoyuan Ma, Guopeng Li, Zheng Wang
Abstract: Text2Motion aims to generate human motions from texts. Existing datasets rely on the assumption that texts include action labels (such as "walk, bend, and pick up"), which is not flexible for practical scenarios. This paper redefines this problem with a more realistic assumption that the texts are arbitrary. Specifically, arbitrary texts include existing action texts composed of action labels (e.g., A person walks and bends to pick up something), and introduce scene texts without explicit action labels (e.g., A person notices his wallet on the ground ahead). To bridge the gaps between this realistic setting and existing datasets, we expand the action texts on the HumanML3D dataset to more scene texts, thereby creating a new HumanML3D++ dataset including arbitrary texts. In this challenging dataset, we benchmark existing state-of-the-art methods and propose a novel two-stage framework to extract action labels from arbitrary texts by the Large Language Model (LLM) and then generate motions from action labels. Extensive experiments are conducted under different application scenarios to validate the effectiveness of the proposed framework on existing and proposed datasets. The results indicate that Text2Motion in this realistic setting is very challenging, fostering new research in this practical direction. Our dataset and code will be released.
Authors: Colby Banbury, Emil Njor, Matthew Stewart, Pete Warden, Manjunath Kudlur, Nat Jeffries, Xenofon Fafoutis, Vijay Janapa Reddi
Abstract: Tiny machine learning (TinyML), which enables machine learning applications on extremely low-power devices, suffers from limited size and quality of relevant datasets. To address this issue, we introduce Wake Vision, a large-scale, diverse dataset tailored for person detection, the canonical task for TinyML visual sensing. Wake Vision comprises over 6 million images, representing a hundredfold increase compared to the previous standard, and has undergone thorough quality filtering. We provide two Wake Vision training sets: Wake Vision (Large) and Wake Vision (Quality), a smaller set with higher-quality labels. Our results demonstrate that using the Wake Vision (Quality) training set produces more accurate models than the Wake Vision (Large) training set, strongly suggesting that label quality is more important than quantity in our setting. We find use for the large training set for pre-training and knowledge distillation. To minimize label errors that can obscure true model performance, we manually label the validation and test sets, improving the test set error rate from 7.8% in the prior standard to only 2.2%. In addition to the dataset, we provide a collection of five detailed benchmark sets to facilitate the evaluation of model quality in challenging real world scenarios that are often ignored when focusing solely on overall accuracy. These novel fine-grained benchmarks assess model performance on specific segments of the test data, such as varying lighting conditions, distances from the camera, and demographic characteristics of subjects. Our results demonstrate that using Wake Vision for training results in a 2.49% increase in accuracy compared to the established dataset. We also show the importance of dataset quality for low-capacity models and the value of dataset size for high-capacity models. wakevision.ai
Authors: Yuan Zhang, Fei Xiao, Tao Huang, Chun-Kai Fan, Hongyuan Dong, Jiawen Li, Jiacong Wang, Kuan Cheng, Shanghang Zhang, Haoyuan Guo
Abstract: Large vision-language models (LVLMs) have recently achieved rapid progress, exhibiting great perception and reasoning abilities concerning visual information. However, when faced with prompts in different sizes of solution spaces, LVLMs fail to always give consistent answers regarding the same knowledge point. This inconsistency of answers between different solution spaces is prevalent in LVLMs and erodes trust. To this end, we provide a multi-modal benchmark ConBench, to intuitively analyze how LVLMs perform when the solution space of a prompt revolves around a knowledge point. Based on the ConBench tool, we are the first to reveal the tapestry and get the following findings: (1) In the discriminate realm, the larger the solution space of the prompt, the lower the accuracy of the answers. (2) Establish the relationship between the discriminative and generative realms: the accuracy of the discriminative question type exhibits a strong positive correlation with its Consistency with the caption. (3) Compared to open-source models, closed-source models exhibit a pronounced bias advantage in terms of Consistency. Eventually, we ameliorate the consistency of LVLMs by trigger-based diagnostic refinement, indirectly improving the performance of their caption. We hope this paper will accelerate the research community in better evaluating their models and encourage future advancements in the consistency domain.
Authors: Xuanjia Zhao, Jian Guan, Congyi Fan, Dongli Xu, Youtian Lin, Haiwei Pan, Pengming Feng
Abstract: Drag-based image editing using generative models provides precise control over image contents, enabling users to manipulate anything in an image with a few clicks. However, prevailing methods typically adopt $n$-step iterations for latent semantic optimization to achieve drag-based image editing, which is time-consuming and limits practical applications. In this paper, we introduce a novel one-step drag-based image editing method, i.e., FastDrag, to accelerate the editing process. Central to our approach is a latent warpage function (LWF), which simulates the behavior of a stretched material to adjust the location of individual pixels within the latent space. This innovation achieves one-step latent semantic optimization and hence significantly promotes editing speeds. Meanwhile, null regions emerging after applying LWF are addressed by our proposed bilateral nearest neighbor interpolation (BNNI) strategy. This strategy interpolates these regions using similar features from neighboring areas, thus enhancing semantic integrity. Additionally, a consistency-preserving strategy is introduced to maintain the consistency between the edited and original images by adopting semantic information from the original image, saved as key and value pairs in self-attention module during diffusion inversion, to guide the diffusion sampling. Our FastDrag is validated on the DragBench dataset, demonstrating substantial improvements in processing time over existing methods, while achieving enhanced editing performance. Project page: https://fastdrag-site.github.io/ .
Authors: Yong Li, Han Gao
Abstract: Neural networks are widely known to be vulnerable to backdoor attacks, a method that poisons a portion of the training data to make the target model perform well on normal data sets, while outputting attacker-specified or random categories on the poisoned samples. Backdoor attacks are full of threats. Poisoned samples are becoming more and more similar to corresponding normal samples, and even the human eye cannot easily distinguish them. On the other hand, the accuracy of models carrying backdoors on normal samples is no different from that of clean models.In this article, by observing the characteristics of backdoor attacks, We provide a new model training method (PT) that freezes part of the model to train a model that can isolate suspicious samples. Then, on this basis, a clean model is fine-tuned to resist backdoor attacks.
Authors: Zhoujie Fu, Jiacheng Wei, Wenhao Shen, Chaoyue Song, Xiaofeng Yang, Fayao Liu, Xulei Yang, Guosheng Lin
Abstract: In this work, we introduce a novel approach for creating controllable dynamics in 3D-generated Gaussians using casually captured reference videos. Our method transfers the motion of objects from reference videos to a variety of generated 3D Gaussians across different categories, ensuring precise and customizable motion transfer. We achieve this by employing blend skinning-based non-parametric shape reconstruction to extract the shape and motion of reference objects. This process involves segmenting the reference objects into motion-related parts based on skinning weights and establishing shape correspondences with generated target shapes. To address shape and temporal inconsistencies prevalent in existing methods, we integrate physical simulation, driving the target shapes with matched motion. This integration is optimized through a displacement loss to ensure reliable and genuine dynamics. Our approach supports diverse reference inputs, including humans, quadrupeds, and articulated objects, and can generate dynamics of arbitrary length, providing enhanced fidelity and applicability. Unlike methods heavily reliant on diffusion video generation models, our technique offers specific and high-quality motion transfer, maintaining both shape integrity and temporal consistency.
Authors: Shenyuan Gao, Jiazhi Yang, Li Chen, Kashyap Chitta, Yihang Qiu, Andreas Geiger, Jun Zhang, Hongyang Li
Abstract: World models can foresee the outcomes of different actions, which is of paramount importance for autonomous driving. Nevertheless, existing driving world models still have limitations in generalization to unseen environments, prediction fidelity of critical details, and action controllability for flexible application. In this paper, we present Vista, a generalizable driving world model with high fidelity and versatile controllability. Based on a systematic diagnosis of existing methods, we introduce several key ingredients to address these limitations. To accurately predict real-world dynamics at high resolution, we propose two novel losses to promote the learning of moving instances and structural information. We also devise an effective latent replacement approach to inject historical frames as priors for coherent long-horizon rollouts. For action controllability, we incorporate a versatile set of controls from high-level intentions (command, goal point) to low-level maneuvers (trajectory, angle, and speed) through an efficient learning strategy. After large-scale training, the capabilities of Vista can seamlessly generalize to different scenarios. Extensive experiments on multiple datasets show that Vista outperforms the most advanced general-purpose video generator in over 70% of comparisons and surpasses the best-performing driving world model by 55% in FID and 27% in FVD. Moreover, for the first time, we utilize the capacity of Vista itself to establish a generalizable reward for real-world action evaluation without accessing the ground truth actions.
Authors: Hanjun Luo, Ziye Deng, Ruizhe Chen, Zuozhu Liu
Abstract: The rapid development and reduced barriers to entry for Text-to-Image (T2I) models have raised concerns about the biases in their outputs, but existing research lacks a holistic definition and evaluation framework of biases, limiting the enhancement of debiasing techniques. To address this issue, we introduce FAIntbench, a holistic and precise benchmark for biases in T2I models. In contrast to existing benchmarks that evaluate bias in limited aspects, FAIntbench evaluate biases from four dimensions: manifestation of bias, visibility of bias, acquired attributes, and protected attributes. We applied FAIntbench to evaluate seven recent large-scale T2I models and conducted human evaluation, whose results demonstrated the effectiveness of FAIntbench in identifying various biases. Our study also revealed new research questions about biases, including the side-effect of distillation. The findings presented here are preliminary, highlighting the potential of FAIntbench to advance future research aimed at mitigating the biases in T2I models. Our benchmark is publicly available to ensure the reproducibility.
Authors: Zixian Guo, Ming Liu, Zhilong Ji, Jinfeng Bai, Yiwen Guo, Wangmeng Zuo
Abstract: Learning a skill generally relies on both practical experience by doer and insightful high-level guidance by instructor. Will this strategy also work well for solving complex non-convex optimization problems? Here, a common gradient-based optimizer acts like a disciplined doer, making locally optimal update at each step. Recent methods utilize large language models (LLMs) to optimize solutions for concrete problems by inferring from natural language instructions, akin to a high-level instructor. In this paper, we show that these two optimizers are complementary to each other, suggesting a collaborative optimization approach. The gradient-based optimizer and LLM-based optimizer are combined in an interleaved manner. We instruct LLMs using task descriptions and timely optimization trajectories recorded during gradient-based optimization. Inferred results from LLMs are used as restarting points for the next stage of gradient optimization. By leveraging both the locally rigorous gradient-based optimizer and the high-level deductive LLM-based optimizer, our combined optimization method consistently yields improvements over competitive baseline prompt tuning methods. Our results demonstrate the synergistic effect of conventional gradient-based optimization and the inference ability of LLMs. The code is released at https://github.com/guozix/LLM-catalyst.
Authors: Yuanjiang Luo, Hongxiang Li, Xuan Wu, Meng Cao, Xiaoshuang Huang, Zhihong Zhu, Peixi Liao, Hu Chen, Yi Zhang
Abstract: Existing mainstream approaches follow the encoder-decoder paradigm for generating radiology reports. They focus on improving the network structure of encoders and decoders, which leads to two shortcomings: overlooking the modality gap and ignoring report content constraints. In this paper, we proposed Textual Inversion and Self-supervised Refinement (TISR) to address the above two issues. Specifically, textual inversion can project text and image into the same space by representing images as pseudo words to eliminate the cross-modeling gap. Subsequently, self-supervised refinement refines these pseudo words through contrastive loss computation between images and texts, enhancing the fidelity of generated reports to images. Notably, TISR is orthogonal to most existing methods, plug-and-play. We conduct experiments on two widely-used public datasets and achieve significant improvements on various baselines, which demonstrates the effectiveness and generalization of TISR. The code will be available soon.
Authors: Khoa Vo, Thinh Phan, Kashu Yamazaki, Minh Tran, Ngan Le
Abstract: Current video-language models (VLMs) rely extensively on instance-level alignment between video and language modalities, which presents two major limitations: (1) visual reasoning disobeys the natural perception that humans do in first-person perspective, leading to a lack of reasoning interpretation; and (2) learning is limited in capturing inherent fine-grained relationships between two modalities. In this paper, we take an inspiration from human perception and explore a compositional approach for egocentric video representation. We introduce HENASY (Hierarchical ENtities ASsemblY), which includes a spatiotemporal token grouping mechanism to explicitly assemble dynamically evolving scene entities through time and model their relationship for video representation. By leveraging compositional structure understanding, HENASY possesses strong interpretability via visual grounding with free-form text queries. We further explore a suite of multi-grained contrastive losses to facilitate entity-centric understandings. This comprises three alignment types: video-narration, noun-entity, verb-entities alignments. Our method demonstrates strong interpretability in both quantitative and qualitative experiments; while maintaining competitive performances on five downstream tasks via zero-shot transfer or as video/text representation, including video/text retrieval, action recognition, multi-choice query, natural language query, and moments query.
Authors: Yunheng Li, ZhongYu Li, Quansheng Zeng, Qibin Hou, Ming-Ming Cheng
Abstract: Pre-trained vision-language models, e.g., CLIP, have been successfully applied to zero-shot semantic segmentation. Existing CLIP-based approaches primarily utilize visual features from the last layer to align with text embeddings, while they neglect the crucial information in intermediate layers that contain rich object details. However, we find that directly aggregating the multi-level visual features weakens the zero-shot ability for novel classes. The large differences between the visual features from different layers make these features hard to align well with the text embeddings. We resolve this problem by introducing a series of independent decoders to align the multi-level visual features with the text embeddings in a cascaded way, forming a novel but simple framework named Cascade-CLIP. Our Cascade-CLIP is flexible and can be easily applied to existing zero-shot semantic segmentation methods. Experimental results show that our simple Cascade-CLIP achieves superior zero-shot performance on segmentation benchmarks, like COCO-Stuff, Pascal-VOC, and Pascal-Context. Our code is available at: https://github.com/HVision-NKU/Cascade-CLIP
Authors: Yuning Zhou, Henry Badgery, Matthew Read, James Bailey, Catherine E. Davey
Abstract: Self-supervised learning (SSL) has potential for effective representation learning in medical imaging, but the choice of data augmentation is critical and domain-specific. It remains uncertain if general augmentation policies suit surgical applications. In this work, we automate the search for suitable augmentation policies through a new method called Dimensionality Driven Augmentation Search (DDA). DDA leverages the local dimensionality of deep representations as a proxy target, and differentiably searches for suitable data augmentation policies in contrastive learning. We demonstrate the effectiveness and efficiency of DDA in navigating a large search space and successfully identifying an appropriate data augmentation policy for laparoscopic surgery. We systematically evaluate DDA across three laparoscopic image classification and segmentation tasks, where it significantly improves over existing baselines. Furthermore, DDA's optimised set of augmentations provides insight into domain-specific dependencies when applying contrastive learning in medical applications. For example, while hue is an effective augmentation for natural images, it is not advantageous for laparoscopic images.
Authors: Enhui Ma, Lijun Zhou, Tao Tang, Zhan Zhang, Dong Han, Junpeng Jiang, Kun Zhan, Peng Jia, Xianpeng Lang, Haiyang Sun, Di Lin, Kaicheng Yu
Abstract: Using generative models to synthesize new data has become a de-facto standard in autonomous driving to address the data scarcity issue. Though existing approaches are able to boost perception models, we discover that these approaches fail to improve the performance of planning of end-to-end autonomous driving models as the generated videos are usually less than 8 frames and the spatial and temporal inconsistencies are not negligible. To this end, we propose Delphi, a novel diffusion-based long video generation method with a shared noise modeling mechanism across the multi-views to increase spatial consistency, and a feature-aligned module to achieves both precise controllability and temporal consistency. Our method can generate up to 40 frames of video without loss of consistency which is about 5 times longer compared with state-of-the-art methods. Instead of randomly generating new data, we further design a sampling policy to let Delphi generate new data that are similar to those failure cases to improve the sample efficiency. This is achieved by building a failure-case driven framework with the help of pre-trained visual language models. Our extensive experiment demonstrates that our Delphi generates a higher quality of long videos surpassing previous state-of-the-art methods. Consequentially, with only generating 4% of the training dataset size, our framework is able to go beyond perception and prediction tasks, for the first time to the best of our knowledge, boost the planning performance of the end-to-end autonomous driving model by a margin of 25%.
Authors: Laura Zheng, Wenjie Wei, Tony Wu, Jacob Clements, Shreelekha Revankar, Andre Harrison, Yu Shen, Ming C. Lin
Abstract: Segmentation is an integral module in many visual computing applications such as virtual try-on, medical imaging, autonomous driving, and agricultural automation. These applications often involve either widespread consumer use or highly variable environments, both of which can degrade the quality of visual sensor data, whether from a common mobile phone or an expensive satellite imaging camera. In addition to external noises like user difference or weather conditions, internal noises such as variations in camera quality or lens distortion can affect the performance of segmentation models during both development and deployment. In this work, we present an efficient, adaptable, and gradient-free method to enhance the robustness of learning-based segmentation models across training. First, we introduce a novel adaptive sensitivity analysis (ASA) using Kernel Inception Distance (KID) on basis perturbations to benchmark perturbation sensitivity of pre-trained segmentation models. Then, we model the sensitivity curve using the adaptive SA and sample perturbation hyperparameter values accordingly. Finally, we conduct adversarial training with the selected perturbation values and dynamically re-evaluate robustness during online training. Our method, implemented end-to-end with minimal fine-tuning required, consistently outperforms state-of-the-art data augmentation techniques for segmentation. It shows significant improvement in both clean data evaluation and real-world adverse scenario evaluation across various segmentation datasets used in visual computing and computer graphics applications.
Authors: Yue Ma, Hongyu Liu, Hongfa Wang, Heng Pan, Yingqing He, Junkun Yuan, Ailing Zeng, Chengfei Cai, Heung-Yeung Shum, Wei Liu, Qifeng Chen
Abstract: We present Follow-Your-Emoji, a diffusion-based framework for portrait animation, which animates a reference portrait with target landmark sequences. The main challenge of portrait animation is to preserve the identity of the reference portrait and transfer the target expression to this portrait while maintaining temporal consistency and fidelity. To address these challenges, Follow-Your-Emoji equipped the powerful Stable Diffusion model with two well-designed technologies. Specifically, we first adopt a new explicit motion signal, namely expression-aware landmark, to guide the animation process. We discover this landmark can not only ensure the accurate motion alignment between the reference portrait and target motion during inference but also increase the ability to portray exaggerated expressions (i.e., large pupil movements) and avoid identity leakage. Then, we propose a facial fine-grained loss to improve the model's ability of subtle expression perception and reference portrait appearance reconstruction by using both expression and facial masks. Accordingly, our method demonstrates significant performance in controlling the expression of freestyle portraits, including real humans, cartoons, sculptures, and even animals. By leveraging a simple and effective progressive generation strategy, we extend our model to stable long-term animation, thus increasing its potential application value. To address the lack of a benchmark for this field, we introduce EmojiBench, a comprehensive benchmark comprising diverse portrait images, driving videos, and landmarks. We show extensive evaluations on EmojiBench to verify the superiority of Follow-Your-Emoji.
Authors: Wenyan Li, Jiaang Li, Rita Ramos, Raphael Tang, Desmond Elliott
Abstract: Recent advances in retrieval-augmented models for image captioning highlight the benefit of retrieving related captions for efficient, lightweight models with strong domain-transfer capabilities. While these models demonstrate the success of retrieval augmentation, retrieval models are still far from perfect in practice: the retrieved information can sometimes mislead the model, resulting in incorrect generation and worse performance. In this paper, we analyze the robustness of a retrieval-augmented captioning model SmallCap. Our analysis shows that the model is sensitive to tokens that appear in the majority of the retrieved captions, and the input attribution shows that those tokens are likely copied into the generated output. Given these findings, we propose to train the model by sampling retrieved captions from more diverse sets. This decreases the chance that the model learns to copy majority tokens, and improves both in-domain and cross-domain performance.
Authors: Clement Chadebec, Onur Tasar, Eyal Benaroche, Benjamin Aubin
Abstract: In this paper, we propose an efficient, fast, and versatile distillation method to accelerate the generation of pre-trained diffusion models: Flash Diffusion. The method reaches state-of-the-art performances in terms of FID and CLIP-Score for few steps image generation on the COCO2014 and COCO2017 datasets, while requiring only several GPU hours of training and fewer trainable parameters than existing methods. In addition to its efficiency, the versatility of the method is also exposed across several tasks such as text-to-image, inpainting, face-swapping, super-resolution and using different backbones such as UNet-based denoisers (SD1.5, SDXL) or DiT (Pixart-$\alpha$), as well as adapters. In all cases, the method allowed to reduce drastically the number of sampling steps while maintaining very high-quality image generation. The official implementation is available at https://github.com/gojasper/flash-diffusion.
Authors: Inkyu Shin, Qihang Yu, Xiaohui Shen, In So Kweon, Kuk-Jin Yoon, Liang-Chieh Chen
Abstract: Recent advancements in zero-shot video diffusion models have shown promise for text-driven video editing, but challenges remain in achieving high temporal consistency. To address this, we introduce Video-3DGS, a 3D Gaussian Splatting (3DGS)-based video refiner designed to enhance temporal consistency in zero-shot video editors. Our approach utilizes a two-stage 3D Gaussian optimizing process tailored for editing dynamic monocular videos. In the first stage, Video-3DGS employs an improved version of COLMAP, referred to as MC-COLMAP, which processes original videos using a Masked and Clipped approach. For each video clip, MC-COLMAP generates the point clouds for dynamic foreground objects and complex backgrounds. These point clouds are utilized to initialize two sets of 3D Gaussians (Frg-3DGS and Bkg-3DGS) aiming to represent foreground and background views. Both foreground and background views are then merged with a 2D learnable parameter map to reconstruct full views. In the second stage, we leverage the reconstruction ability developed in the first stage to impose the temporal constraints on the video diffusion model. To demonstrate the efficacy of Video-3DGS on both stages, we conduct extensive experiments across two related tasks: Video Reconstruction and Video Editing. Video-3DGS trained with 3k iterations significantly improves video reconstruction quality (+3 PSNR, +7 PSNR increase) and training efficiency (x1.9, x4.5 times faster) over NeRF-based and 3DGS-based state-of-art methods on DAVIS dataset, respectively. Moreover, it enhances video editing by ensuring temporal consistency across 58 dynamic monocular videos.
Authors: Peng Xing, Ning Wang, Jianbo Ouyang, Zechao Li
Abstract: The remarkable advancement in text-to-image generation models significantly boosts the research in ID customization generation. However, existing personalization methods cannot simultaneously satisfy high fidelity and high-efficiency requirements. Their main bottleneck lies in the prompt image encoder, which produces weak alignment signals with the text-to-image model and significantly increased model size. Towards this end, we propose a lightweight Inv-Adapter, which first extracts diffusion-domain representations of ID images utilizing a pre-trained text-to-image model via DDIM image inversion, without additional image encoder. Benefiting from the high alignment of the extracted ID prompt features and the intermediate features of the text-to-image model, we then embed them efficiently into the base text-to-image model by carefully designing a lightweight attention adapter. We conduct extensive experiments to assess ID fidelity, generation loyalty, speed, and training parameters, all of which show that the proposed Inv-Adapter is highly competitive in ID customization generation and model scale.
Authors: Minglei Li, Peng Ye, Yongqi Huang, Lin Zhang, Tao Chen, Tong He, Jiayuan Fan, Wanli Ouyang
Abstract: Parameter-efficient fine-tuning (PEFT) has become increasingly important as foundation models continue to grow in both popularity and size. Adapter has been particularly well-received due to their potential for parameter reduction and adaptability across diverse tasks. However, striking a balance between high efficiency and robust generalization across tasks remains a challenge for adapter-based methods. We analyze existing methods and find that: 1) parameter sharing is the key to reducing redundancy; 2) more tunable parameters, dynamic allocation, and block-specific design are keys to improving performance. Unfortunately, no previous work considers all these factors. Inspired by this insight, we introduce a novel framework named Adapter-X. First, a Sharing Mixture of Adapters (SMoA) module is proposed to fulfill token-level dynamic allocation, increased tunable parameters, and inter-block sharing at the same time. Second, some block-specific designs like Prompt Generator (PG) are introduced to further enhance the ability of adaptation. Extensive experiments across 2D image and 3D point cloud modalities demonstrate that Adapter-X represents a significant milestone as it is the first to outperform full fine-tuning in both 2D image and 3D point cloud modalities with significantly fewer parameters, i.e., only 0.20% and 1.88% of original trainable parameters for 2D and 3D classification tasks. Our code will be publicly available.
Authors: Ryo Fujii, Hideo Saito, Hiroki Kajita
Abstract: Surgical tool detection is a fundamental task for understanding egocentric open surgery videos. However, detecting surgical tools presents significant challenges due to their highly imbalanced class distribution, similar shapes and similar textures, and heavy occlusion. The lack of a comprehensive large-scale dataset compounds these challenges. In this paper, we introduce EgoSurgery-Tool, an extension of the existing EgoSurgery-Phase dataset, which contains real open surgery videos captured using an egocentric camera attached to the surgeon's head, along with phase annotations. EgoSurgery-Tool has been densely annotated with surgical tools and comprises over 49K surgical tool bounding boxes across 15 categories, constituting a large-scale surgical tool detection dataset. EgoSurgery-Tool also provides annotations for hand detection with over 46K hand-bounding boxes, capturing hand-object interactions that are crucial for understanding activities in egocentric open surgery. EgoSurgery-Tool is superior to existing datasets due to its larger scale, greater variety of surgical tools, more annotations, and denser scenes. We conduct a comprehensive analysis of EgoSurgery-Tool using nine popular object detectors to assess their effectiveness in both surgical tool and hand detection. The dataset will be released at https://github.com/Fujiry0/EgoSurgery.
Authors: Jiangning Zhang, Haoyang He, Zhenye Gan, Qingdong He, Yuxuan Cai, Zhucun Xue, Yabiao Wang, Chengjie Wang, Lei Xie, Yong Liu
Abstract: Visual anomaly detection aims to identify anomalous regions in images through unsupervised learning paradigms, with increasing application demand and value in fields such as industrial inspection and medical lesion detection. Despite significant progress in recent years, there is a lack of comprehensive benchmarks to adequately evaluate the performance of various mainstream methods across different datasets under the practical multi-class setting. The absence of standardized experimental setups can lead to potential biases in training epochs, resolution, and metric results, resulting in erroneous conclusions. This paper addresses this issue by proposing a comprehensive visual anomaly detection benchmark, \textbf{\textit{ADer}}, which is a modular framework that is highly extensible for new methods. The benchmark includes multiple datasets from industrial and medical domains, implementing fifteen state-of-the-art methods and nine comprehensive metrics. Additionally, we have open-sourced the GPU-assisted \href{https://pypi.org/project/ADEval}{ADEval} package to address the slow evaluation problem of metrics like time-consuming mAU-PRO on large-scale data, significantly reducing evaluation time by more than \textit{1000-fold}. Through extensive experimental results, we objectively reveal the strengths and weaknesses of different methods and provide insights into the challenges and future directions of multi-class visual anomaly detection. We hope that \textbf{\textit{ADer}} will become a valuable resource for researchers and practitioners in the field, promoting the development of more robust and generalizable anomaly detection systems. Full codes have been attached in Appendix and open-sourced at \url{https://github.com/zhangzjn/ader}.
URLs: https://pypi.org/project/ADEval, https://github.com/zhangzjn/ader
Authors: Kailas Vodrahalli, James Zou
Abstract: As generative AI becomes more prevalent, it is important to study how human users interact with such models. In this work, we investigate how people use text-to-image models to generate desired target images. To study this interaction, we created ArtWhisperer, an online game where users are given a target image and are tasked with iteratively finding a prompt that creates a similar-looking image as the target. Through this game, we recorded over 50,000 human-AI interactions; each interaction corresponds to one text prompt created by a user and the corresponding generated image. The majority of these are repeated interactions where a user iterates to find the best prompt for their target image, making this a unique sequential dataset for studying human-AI collaborations. In an initial analysis of this dataset, we identify several characteristics of prompt interactions and user strategies. People submit diverse prompts and are able to discover a variety of text descriptions that generate similar images. Interestingly, prompt diversity does not decrease as users find better prompts. We further propose a new metric to quantify the steerability of AI using our dataset. We define steerability as the expected number of interactions required to adequately complete a task. We estimate this value by fitting a Markov chain for each target task and calculating the expected time to reach an adequate score in the Markov chain. We quantify and compare AI steerability across different types of target images and two different models, finding that images of cities and natural world images are more steerable than artistic and fantasy images. These findings provide insights into human-AI interaction behavior, present a concrete method of assessing AI steerability, and demonstrate the general utility of the ArtWhisperer dataset.
Authors: Jingwen Fu, Tao Yang, Yuwang Wang, Yan Lu, Nanning Zheng
Abstract: In-context learning, i.e., learning from context examples, is an impressive ability of Transformer. Training Transformers to possess this in-context learning skill is computationally intensive due to the occurrence of learning plateaus, which are periods within the training process where there is minimal or no enhancement in the model's in-context learning capability. To study the mechanism behind the learning plateaus, we conceptually seperate a component within the model's internal representation that is exclusively affected by the model's weights. We call this the "weights component", and the remainder is identified as the "context component". By conducting meticulous and controlled experiments on synthetic tasks, we note that the persistence of learning plateaus correlates with compromised functionality of the weights component. Recognizing the impaired performance of the weights component as a fundamental behavior drives learning plateaus, we have developed three strategies to expedite the learning of Transformers. The effectiveness of these strategies is further confirmed in natural language processing tasks. In conclusion, our research demonstrates the feasibility of cultivating a powerful in-context learning ability within AI systems in an eco-friendly manner.
Authors: Pascal Peter
Abstract: Diffusion probabilistic models excel at sampling new images from learned distributions. Originally motivated by drift-diffusion concepts from physics, they apply image perturbations such as noise and blur in a forward process that results in a tractable probability distribution. A corresponding learned reverse process generates images and can be conditioned on side information, which leads to a wide variety of practical applications. Most of the research focus currently lies on practice-oriented extensions. In contrast, the theoretical background remains largely unexplored, in particular the relations to drift-diffusion. In order to shed light on these connections to classical image filtering, we propose a generalised scale-space theory for diffusion probabilistic models. Moreover, we show conceptual and empirical connections to diffusion and osmosis filters.
Authors: Anay Majee, Suraj Kothawade, Krishnateja Killamsetty, Rishabh Iyer
Abstract: In this paper we introduce the SCoRe (Submodular Combinatorial Representation Learning) framework, a novel approach in representation learning that addresses inter-class bias and intra-class variance. SCoRe provides a new combinatorial viewpoint to representation learning, by introducing a family of loss functions based on set-based submodular information measures. We develop two novel combinatorial formulations for loss functions, using the Total Information and Total Correlation, that naturally minimize intra-class variance and inter-class bias. Several commonly used metric/contrastive learning loss functions like supervised contrastive loss, orthogonal projection loss, and N-pairs loss, are all instances of SCoRe, thereby underlining the versatility and applicability of SCoRe in a broad spectrum of learning scenarios. Novel objectives in SCoRe naturally model class-imbalance with up to 7.6\% improvement in classification on CIFAR-10-LT, CIFAR-100-LT, MedMNIST, 2.1% on ImageNet-LT, and 19.4% in object detection on IDD and LVIS (v1.0), demonstrating its effectiveness over existing approaches.
Authors: Andy Zhou, Kai Yan, Michal Shlapentokh-Rothman, Haohan Wang, Yu-Xiong Wang
Abstract: While language models (LMs) have shown potential across a range of decision-making tasks, their reliance on simple acting processes limits their broad deployment as autonomous agents. In this paper, we introduce Language Agent Tree Search (LATS) -- the first general framework that synergizes the capabilities of LMs in reasoning, acting, and planning. By leveraging the in-context learning ability of LMs, we integrate Monte Carlo Tree Search into LATS to enable LMs as agents, along with LM-powered value functions and self-reflections for proficient exploration and enhanced decision-making. A key feature of our approach is the incorporation of an environment for external feedback, which offers a more deliberate and adaptive problem-solving mechanism that surpasses the constraints of existing techniques. Our experimental evaluation across diverse domains, including programming, interactive question-answering (QA), web navigation, and math, validates the effectiveness and generality of LATS in decision-making while maintaining competitive or improved reasoning performance. Notably, LATS achieves state-of-the-art pass@1 accuracy (92.7%) for programming on HumanEval with GPT-4 and demonstrates gradient-free performance (average score of 75.9) comparable to gradient-based fine-tuning for web navigation on WebShop with GPT-3.5. Code can be found at https://github.com/lapisrocks/LanguageAgentTreeSearch
Authors: Jianguo Huang, Huajun Xi, Linjun Zhang, Huaxiu Yao, Yue Qiu, Hongxin Wei
Abstract: Conformal prediction is a statistical framework that generates prediction sets containing ground-truth labels with a desired coverage guarantee. The predicted probabilities produced by machine learning models are generally miscalibrated, leading to large prediction sets in conformal prediction. To address this issue, we propose a novel algorithm named $\textit{Sorted Adaptive Prediction Sets}$ (SAPS), which discards all the probability values except for the maximum softmax probability. The key idea behind SAPS is to minimize the dependence of the non-conformity score on the probability values while retaining the uncertainty information. In this manner, SAPS can produce compact prediction sets and communicate instance-wise uncertainty. Extensive experiments validate that SAPS not only lessens the prediction sets but also broadly enhances the conditional coverage rate of prediction sets.
Authors: David T. Hoffmann, Simon Schrodi, Jelena Bratuli\'c, Nadine Behrmann, Volker Fischer, Thomas Brox
Abstract: In this work, we study rapid improvements of the training loss in transformers when being confronted with multi-step decision tasks. We found that transformers struggle to learn the intermediate task and both training and validation loss saturate for hundreds of epochs. When transformers finally learn the intermediate task, they do this rapidly and unexpectedly. We call these abrupt improvements Eureka-moments, since the transformer appears to suddenly learn a previously incomprehensible concept. We designed synthetic tasks to study the problem in detail, but the leaps in performance can be observed also for language modeling and in-context learning (ICL). We suspect that these abrupt transitions are caused by the multi-step nature of these tasks. Indeed, we find connections and show that ways to improve on the synthetic multi-step tasks can be used to improve the training of language modeling and ICL. Using the synthetic data we trace the problem back to the Softmax function in the self-attention block of transformers and show ways to alleviate the problem. These fixes reduce the required number of training steps, lead to higher likelihood to learn the intermediate task, to higher final accuracy and training becomes more robust to hyper-parameters.
Authors: Junli Fang, Jo\~ao F. C. Mota, Baoshan Lu, Weicheng Zhang, Xuemin Hong
Abstract: The joint source-channel coding (JSCC) framework leverages deep learning to learn from data the best codes for source and channel coding. When the output signal, rather than being binary, is directly mapped onto the IQ domain (complex-valued), we call the resulting framework joint source coding and modulation (JSCM). We consider a JSCM scenario and show the existence of a strict tradeoff between channel rate, distortion, perception, and classification accuracy, a tradeoff that we name RDPC. We then propose two image compression methods to navigate that tradeoff: the RDPCO algorithm which, under simple assumptions, directly solves the optimization problem characterizing the tradeoff, and an algorithm based on an inverse-domain generative adversarial network (ID-GAN), which is more general and achieves extreme compression. Simulation results corroborate the theoretical findings, showing that both algorithms exhibit the RDPC tradeoff. They also demonstrate that the proposed ID-GAN algorithm effectively balances image distortion, perception, and classification accuracy, and significantly outperforms traditional separation-based methods and recent deep JSCM architectures in terms of one or more of these metrics.
Authors: Jing Yu Koh, Robert Lo, Lawrence Jang, Vikram Duvvur, Ming Chong Lim, Po-Yu Huang, Graham Neubig, Shuyan Zhou, Ruslan Salakhutdinov, Daniel Fried
Abstract: Autonomous agents capable of planning, reasoning, and executing actions on the web offer a promising avenue for automating computer tasks. However, the majority of existing benchmarks primarily focus on text-based agents, neglecting many natural tasks that require visual information to effectively solve. Given that most computer interfaces cater to human perception, visual information often augments textual data in ways that text-only models struggle to harness effectively. To bridge this gap, we introduce VisualWebArena, a benchmark designed to assess the performance of multimodal web agents on realistic \textit{visually grounded tasks}. VisualWebArena comprises of a set of diverse and complex web-based tasks that evaluate various capabilities of autonomous multimodal agents. To perform on this benchmark, agents need to accurately process image-text inputs, interpret natural language instructions, and execute actions on websites to accomplish user-defined objectives. We conduct an extensive evaluation of state-of-the-art LLM-based autonomous agents, including several multimodal models. Through extensive quantitative and qualitative analysis, we identify several limitations of text-only LLM agents, and reveal gaps in the capabilities of state-of-the-art multimodal language agents. VisualWebArena provides a framework for evaluating multimodal autonomous language agents, and offers insights towards building stronger autonomous agents for the web. Our code, baseline models, and data is publicly available at https://jykoh.com/vwa.
URLs: https://jykoh.com/vwa.
Authors: Andy Zhou, Bo Li, Haohan Wang
Abstract: Despite advances in AI alignment, large language models (LLMs) remain vulnerable to adversarial attacks or jailbreaking, in which adversaries can modify prompts to induce unwanted behavior. While some defenses have been proposed, they have not been adapted to newly proposed attacks and more challenging threat models. To address this, we propose an optimization-based objective for defending LLMs against jailbreaking attacks and an algorithm, Robust Prompt Optimization (RPO) to create robust system-level defenses. Our approach directly incorporates the adversary into the defensive objective and optimizes a lightweight and transferable suffix, enabling RPO to adapt to worst-case adaptive attacks. Our theoretical and experimental results show improved robustness to both jailbreaks seen during optimization and unknown jailbreaks, reducing the attack success rate (ASR) on GPT-4 to 6% and Llama-2 to 0% on JailbreakBench, setting the state-of-the-art. Code can be found at https://github.com/lapisrocks/rpo
Authors: Haoyi Zhu, Yating Wang, Di Huang, Weicai Ye, Wanli Ouyang, Tong He
Abstract: In robot learning, the observation space is crucial due to the distinct characteristics of different modalities, which can potentially become a bottleneck alongside policy design. In this study, we explore the influence of various observation spaces on robot learning, focusing on three predominant modalities: RGB, RGB-D, and point cloud. We introduce OBSBench, a benchmark comprising two simulators and 125 tasks, along with standardized pipelines for various encoders and policy baselines. Extensive experiments on diverse contact-rich manipulation tasks reveal a notable trend: point cloud-based methods, even those with the simplest designs, frequently outperform their RGB and RGB-D counterparts. This trend persists in both scenarios: training from scratch and utilizing pre-training. Furthermore, our findings demonstrate that point cloud observations often yield better policy performance and significantly stronger generalization capabilities across various geometric and visual conditions. These outcomes suggest that the 3D point cloud is a valuable observation modality for intricate robotic tasks. We also suggest that incorporating both appearance and coordinate information can enhance the performance of point cloud methods. We hope our work provides valuable insights and guidance for designing more generalizable and robust robotic models. Codes are available at https://github.com/HaoyiZhu/PointCloudMatters.
Authors: Eduard Zamfir, Zongwei Wu, Nancy Mehta, Yulun Zhang, Radu Timofte
Abstract: Reconstructing high-resolution (HR) images from low-resolution (LR) inputs poses a significant challenge in image super-resolution (SR). While recent approaches have demonstrated the efficacy of intricate operations customized for various objectives, the straightforward stacking of these disparate operations can result in a substantial computational burden, hampering their practical utility. In response, we introduce SeemoRe, an efficient SR model employing expert mining. Our approach strategically incorporates experts at different levels, adopting a collaborative methodology. At the macro scale, our experts address rank-wise and spatial-wise informative features, providing a holistic understanding. Subsequently, the model delves into the subtleties of rank choice by leveraging a mixture of low-rank experts. By tapping into experts specialized in distinct key factors crucial for accurate SR, our model excels in uncovering intricate intra-feature details. This collaborative approach is reminiscent of the concept of "see more", allowing our model to achieve an optimal performance with minimal computational costs in efficient settings. The source will be publicly made available at https://github.com/eduardzamfir/seemoredetails
Authors: Canyu Zhang, Youbao Tang, Ning Zhang, Ruei-Sung Lin, Mei Han, Jing Xiao, Song Wang
Abstract: Dance serves as a powerful medium for expressing human emotions, but the lifelike generation of dance is still a considerable challenge. Recently, diffusion models have showcased remarkable generative abilities across various domains. They hold promise for human motion generation due to their adaptable many-to-many nature. Nonetheless, current diffusion-based motion generation models often create entire motion sequences directly and unidirectionally, lacking focus on the motion with local and bidirectional enhancement. When choreographing high-quality dance movements, people need to take into account not only the musical context but also the nearby music-aligned dance motions. To authentically capture human behavior, we propose a Bidirectional Autoregressive Diffusion Model (BADM) for music-to-dance generation, where a bidirectional encoder is built to enforce that the generated dance is harmonious in both the forward and backward directions. To make the generated dance motion smoother, a local information decoder is built for local motion enhancement. The proposed framework is able to generate new motions based on the input conditions and nearby motions, which foresees individual motion slices iteratively and consolidates all predictions. To further refine the synchronicity between the generated dance and the beat, the beat information is incorporated as an input to generate better music-aligned dance movements. Experimental results demonstrate that the proposed model achieves state-of-the-art performance compared to existing unidirectional approaches on the prominent benchmark for music-to-dance generation.
Authors: Dongping Chen, Ruoxi Chen, Shilin Zhang, Yinuo Liu, Yaochen Wang, Huichi Zhou, Qihui Zhang, Pan Zhou, Yao Wan, Lichao Sun
Abstract: Multimodal Large Language Models (MLLMs) have gained significant attention recently, showing remarkable potential in artificial general intelligence. However, assessing the utility of MLLMs presents considerable challenges, primarily due to the absence of multimodal benchmarks that align with human preferences. Drawing inspiration from the concept of LLM-as-a-Judge within LLMs, this paper introduces a novel benchmark, termed MLLM-as-a-Judge, to assess the ability of MLLMs in assisting judges across diverse modalities, encompassing three distinct tasks: Scoring Evaluation, Pair Comparison, and Batch Ranking. Our study reveals that, while MLLMs demonstrate remarkable human-like discernment in Pair Comparison, there is a significant divergence from human preferences in Scoring Evaluation and Batch Ranking. Furthermore, a closer examination reveals persistent challenges in the judgment capacities of LLMs, including diverse biases, hallucinatory responses, and inconsistencies in judgment, even in advanced models such as GPT-4V. These findings emphasize the pressing need for enhancements and further research efforts to be undertaken before regarding MLLMs as fully reliable evaluators. In light of this, we advocate for additional efforts dedicated to supporting the continuous development within the domain of MLLM functioning as judges. The code and dataset are publicly available at our project homepage: \url{https://mllm-judge.github.io/}.
Authors: Naihao Deng, Zhenjie Sun, Ruiqi He, Aman Sikka, Yulong Chen, Lin Ma, Yue Zhang, Rada Mihalcea
Abstract: In this paper, we investigate the effectiveness of various LLMs in interpreting tabular data through different prompting strategies and data formats. Our analyses extend across six benchmarks for table-related tasks such as question-answering and fact-checking. We introduce for the first time the assessment of LLMs' performance on image-based table representations. Specifically, we compare five text-based and three image-based table representations, demonstrating the role of representation and prompting on LLM performance. Our study provides insights into the effective use of LLMs on table-related tasks.
Authors: Haoyi Qiu, Wenbo Hu, Zi-Yi Dou, Nanyun Peng
Abstract: Large Vision-Language Models (LVLMs) suffer from hallucination issues, wherein the models generate plausible-sounding but factually incorrect outputs, undermining their reliability. A comprehensive quantitative evaluation is necessary to identify and understand the extent of hallucinations in these models. However, existing benchmarks are often limited in scope, focusing mainly on object hallucinations. Furthermore, current evaluation methods struggle to effectively address the subtle semantic distinctions between model outputs and reference data, as well as the balance between hallucination and informativeness. To address these issues, we introduce a multi-dimensional benchmark covering objects, attributes, and relations, with challenging images selected based on associative biases. Moreover, we propose a large language model (LLM)-based two-stage evaluation framework that generalizes the popular CHAIR metric and incorporates both faithfulness and coverage into the evaluation. Experiments on 10 established LVLMs demonstrate that our evaluation metric is more comprehensive and better correlated with humans than existing work when evaluating on our challenging human-annotated benchmark dataset. Our work also highlights the critical balance between faithfulness and coverage of model outputs, and encourages future works to address hallucinations in LVLMs while keeping their outputs informative.
Authors: Andrew Kyle Lampinen, Stephanie C. Y. Chan, Katherine Hermann
Abstract: Representation learning, and interpreting learned representations, are key areas of focus in machine learning and neuroscience. Both fields generally use representations as a means to understand or improve a system's computations. In this work, however, we explore surprising dissociations between representation and computation that may pose challenges for such efforts. We create datasets in which we attempt to match the computational role that different features play, while manipulating other properties of the features or the data. We train various deep learning architectures to compute these multiple abstract features about their inputs. We find that their learned feature representations are systematically biased towards representing some features more strongly than others, depending upon extraneous properties such as feature complexity, the order in which features are learned, and the distribution of features over the inputs. For example, features that are simpler to compute or learned first tend to be represented more strongly and densely than features that are more complex or learned later, even if all features are learned equally well. We also explore how these biases are affected by architectures, optimizers, and training regimes (e.g., in transformers, features decoded earlier in the output sequence also tend to be represented more strongly). Our results help to characterize the inductive biases of gradient-based representation learning. These results also highlight a key challenge for interpretability $-$ or for comparing the representations of models and brains $-$ disentangling extraneous biases from the computationally important aspects of a system's internal representations.
Authors: Bowen Jiang, Yangxinyu Xie, Xiaomeng Wang, Weijie J. Su, Camillo J. Taylor, Tanwi Mallick
Abstract: Rationality is the quality of being guided by reason, characterized by logical thinking and decision-making that align with evidence and logical rules. This quality is essential for effective problem-solving, as it ensures that solutions are well-founded and systematically derived. Despite the advancements of large language models (LLMs) in generating human-like text with remarkable accuracy, they present biases inherited from the training data, inconsistency across different contexts, and difficulty understanding complex scenarios involving multiple layers of context. Therefore, recent research attempts to leverage the strength of multiple agents working collaboratively with various types of data and tools for enhanced consistency and reliability. To that end, this paper aims to understand whether multi-modal and multi-agent systems are advancing toward rationality by surveying the state-of-the-art works, identifying advancements over single-agent and single-modal systems in terms of rationality, and discussing open problems and future directions. We maintain an open repository at https://github.com/bowen-upenn/MMMA_Rationality.
Authors: Yundi Zhang, Chen Chen, Suprosanna Shit, Sophie Starck, Daniel Rueckert, Jiazhen Pan
Abstract: Cardiac Magnetic Resonance (CMR) imaging serves as the gold-standard for evaluating cardiac morphology and function. Typically, a multi-view CMR stack, covering short-axis (SA) and 2/3/4-chamber long-axis (LA) views, is acquired for a thorough cardiac assessment. However, efficiently streamlining the complex, high-dimensional 3D+T CMR data and distilling compact, coherent representation remains a challenge. In this work, we introduce a whole-heart self-supervised learning framework that utilizes masked imaging modeling to automatically uncover the correlations between spatial and temporal patches throughout the cardiac stacks. This process facilitates the generation of meaningful and well-clustered heart representations without relying on the traditionally required, and often costly, labeled data. The learned heart representation can be directly used for various downstream tasks. Furthermore, our method demonstrates remarkable robustness, ensuring consistent representations even when certain CMR planes are missing/flawed. We train our model on 14,000 unlabeled CMR data from UK BioBank and evaluate it on 1,000 annotated data. The proposed method demonstrates superior performance to baselines in tasks that demand comprehensive 3D+T cardiac information, e.g. cardiac phenotype (ejection fraction and ventricle volume) prediction and multi-plane/multi-frame CMR segmentation, highlighting its effectiveness in extracting comprehensive cardiac features that are both anatomically and pathologically relevant.
Authors: Jincheng Zhong, Xingzhuo Guo, Jiaxiang Dong, Mingsheng Long
Abstract: Diffusion models have significantly advanced the field of generative modeling. However, training a diffusion model is computationally expensive, creating a pressing need to adapt off-the-shelf diffusion models for downstream generation tasks. Current fine-tuning methods focus on parameter-efficient transfer learning but overlook the fundamental transfer characteristics of diffusion models. In this paper, we investigate the transferability of diffusion models and observe a monotonous chain of forgetting trend of transferability along the reverse process. Based on this observation and novel theoretical insights, we present Diff-Tuning, a frustratingly simple transfer approach that leverages the chain of forgetting tendency. Diff-Tuning encourages the fine-tuned model to retain the pre-trained knowledge at the end of the denoising chain close to the generated data while discarding the other noise side. We conduct comprehensive experiments to evaluate Diff-Tuning, including the transfer of pre-trained Diffusion Transformer models to eight downstream generations and the adaptation of Stable Diffusion to five control conditions with ControlNet. Diff-Tuning achieves a 26% improvement over standard fine-tuning and enhances the convergence speed of ControlNet by 24%. Notably, parameter-efficient transfer learning techniques for diffusion models can also benefit from Diff-Tuning.
Authors: Jifei Luo, Hantao Yao, Changsheng Xu
Abstract: Diffusion-based re-ranking is a common method used for retrieving instances by performing similarity propagation in a nearest neighbor graph. However, existing techniques that construct the affinity graph based on pairwise instances can lead to the propagation of misinformation from outliers and other manifolds, resulting in inaccurate results. To overcome this issue, we propose a novel Cluster-Aware Similarity (CAS) diffusion for instance retrieval. The primary concept of CAS is to conduct similarity diffusion within local clusters, which can reduce the influence from other manifolds explicitly. To obtain a symmetrical and smooth similarity matrix, our Bidirectional Similarity Diffusion strategy introduces an inverse constraint term to the optimization objective of local cluster diffusion. Additionally, we have optimized a Neighbor-guided Similarity Smoothing approach to ensure similarity consistency among the local neighbors of each instance. Evaluations in instance retrieval and object re-identification validate the effectiveness of the proposed CAS, our code is publicly available.
Authors: Chenxin Li, Xinyu Liu, Wuyang Li, Cheng Wang, Hengyu Liu, Yixuan Yuan
Abstract: U-Net has become a cornerstone in various visual applications such as image segmentation and diffusion probability models. While numerous innovative designs and improvements have been introduced by incorporating transformers or MLPs, the networks are still limited to linearly modeling patterns as well as the deficient interpretability. To address these challenges, our intuition is inspired by the impressive results of the Kolmogorov-Arnold Networks (KANs) in terms of accuracy and interpretability, which reshape the neural network learning via the stack of non-linear learnable activation functions derived from the Kolmogorov-Anold representation theorem. Specifically, in this paper, we explore the untapped potential of KANs in improving backbones for vision tasks. We investigate, modify and re-design the established U-Net pipeline by integrating the dedicated KAN layers on the tokenized intermediate representation, termed U-KAN. Rigorous medical image segmentation benchmarks verify the superiority of U-KAN by higher accuracy even with less computation cost. We further delved into the potential of U-KAN as an alternative U-Net noise predictor in diffusion models, demonstrating its applicability in generating task-oriented model architectures. These endeavours unveil valuable insights and sheds light on the prospect that with U-KAN, you can make strong backbone for medical image segmentation and generation. Project page: https://yes-ukan.github.io/