We propose a new formulation of temporal action detection (TAD) with denoising diffusion, DiffTAD in short. Taking as input random temporal proposals, it can yield action proposals accurately given an untrimmed long video. This presents a generative modeling perspective, against previous discriminative learning manners. This capability is achieved by first diffusing the ground-truth proposals to random ones (i.e, the forward/noising process) and then learning to reverse the noising process (i.e, the backward/denoising process). Concretely, we establish the denoising process in the Transformer decoder (e.g, DETR) by introducing a temporal location query design with faster convergence in training. We further propose a cross-step selective conditioning algorithm for inference acceleration. Extensive evaluations on ActivityNet and THUMOS show that our DiffTAD achieves top performance compared to previous art alternatives. The code is available at https://github.com/sauradip/DiffusionTAD.
论文检索
输入标题、作者或关键词，从 2,156 篇学术成果中精准定位
Taxonomy Adaptive Cross-Domain Adaptation in Medical Imaging via Optimization Trajectory Distillation
PDF ↗The success of automated medical image analysis depends on large-scale and expert-annotated training sets. Unsupervised domain adaptation (UDA) has been raised as a promising approach to alleviate the burden of labeled data collection. However, they generally operate under the closed-set adaptation setting assuming an identical label set between the source and target domains, which is over-restrictive in clinical practice where new classes commonly exist across datasets due to taxonomic inconsistency. While several methods have been presented to tackle both domain shifts and incoherent label sets, none of them take into account the common characteristics of the two issues and consider the learning dynamics along network training. In this work, we propose optimization trajectory distillation, a unified approach to address the two technical challenges from a new perspective. It exploits the low-rank nature of gradient space and devises a dual-stream distillation algorithm to regularize the learning dynamics of insufficiently annotated domain and classes with the external guidance obtained from reliable sources. Our approach resolves the issue of inadequate navigation along network optimization, which is the major obstacle in the taxonomy adaptive cross-domain adaptation scenario. We evaluate the proposed method extensively on several tasks towards various endpoints with clinical significance. The results demonstrate its effectiveness and improvements over previous methods.
Towards Deeply Unified Depth-aware Panoptic Segmentation with Bi-directional Guidance Learning
PDF ↗Depth-aware panoptic segmentation is an emerging topic in computer vision which combines semantic and geometric understanding for more robust scene interpretation. Recent works pursue unified frameworks to tackle this challenge but mostly still treat it as two individual learning tasks, which limits their potential for exploring cross-domain information. We propose a deeply unified framework for depth-aware panoptic segmentation, which performs joint segmentation and depth estimation both in a per-segment manner with identical object queries. To narrow the gap between the two tasks, we further design a geometric query enhancement method, which is able to integrate scene geometry into object queries using latent representations. In addition, we propose a bi-directional guidance learning approach to facilitate cross-task feature learning by taking advantage of their mutual relations. Our method sets the new state of the art for depth-aware panoptic segmentation on both Cityscapes-DVPS and SemKITTI-DVPS datasets. Moreover, our guidance learning approach is shown to deliver performance improvement even under incomplete supervision labels. Code and models are available at https://github.com/jwh97nn/DeepDPS.
In this work, we investigate the problem of creating high-fidelity 3D content from only a single image. This is inherently challenging: it essentially involves estimating the underlying 3D geometry while hallucinating unseen textures. To address this challenge, we leverage prior knowledge in a well-trained 2D diffusion model to serve as a 3D-aware supervision for 3D creation. Our proposed method, Make-It-3D, employs a two-stage optimization pipeline: the first stage optimizes a neural radiance field with constraints from the reference image and diffusion prior; the second stage builds textured point clouds from the coarse model and further enhances the textures with diffusion prior leveraging the availability of high-quality textures from the reference image. Extensive experiments show that our method achieves a clear improvement over previous works, displaying faithful reconstruction and impressive visual quality. Our method presents the first attempt to achieve high-quality 3D creation from a single image for general objects, and enables various applications such as text-to-3D creation and texture editing.
This paper examines the problems of severe image-text misalignment and high redundancy in the widely-used large-scale Vision-Language Pre-Training (VLP) datasets. To address these issues, we propose an efficient and straightforward Vision-Language learning algorithm called TL;DR which aims to compress the existing large VLP data into a small, high-quality set. Our approach consists of two major steps. First, a codebook-based encoder-decoder captioner is developed to select representative samples. Second, a new caption is generated to complement the original captions for selected samples, mitigating the text-image misalignment problem while maintaining uniqueness. As the result, TL;DR enables us to reduce the large dataset into a small set of high-quality data, which can serve as an alternative pre-training dataset. This algorithm significantly speeds up the time-consuming pretraining process. Specifically, TL;DR can compress the mainstream VLP datasets at a high ratio, e.g., reduce well-cleaned CC3M dataset from 2.8M to 0.67M ( 24%) and noisy YFCC15M from 15M to 2.5M ( 16.7%). Extensive experiments with three popular VLP models over seven downstream tasks show that VLP model trained on the compressed dataset provided by TL;DR can perform similar or even better results compared with training on the full-scale dataset.
Multi-Camera Multi-Object Tracking (MC-MOT) utilizes information from multiple views to better handle problems with occlusion and crowded scenes. Recently, the use of graph-based approaches to solve tracking problems has become very popular. However, many current graph-based methods do not effectively utilize information regarding spatial and temporal consistency. Instead, they rely on single-camera trackers as input, which are prone to fragmentation and ID switch errors. In this paper, we propose a novel reconfigurable graph model that first associates all detected objects across cameras spatially before reconfiguring it into a temporal graph for Temporal Association. This two-stage association approach enables us to extract robust spatial and temporal-aware features and address the problem with fragmented tracklets. Furthermore, our model is designed for online tracking, making it suitable for real-world applications. Experimental results show that the proposed graph model is able to extract more discriminating features for object tracking, and our model achieves state-of-the-art performance on several public datasets. Code is available at https://github.com/chengche6230/ReST.
This study explores the concept of equivariance in vision-language foundation models (VLMs), focusing specifically on the multimodal similarity function that is not only the major training objective but also the core delivery to support downstream tasks. Unlike the existing image-text similarity objective which only categorizes matched pairs as similar and unmatched as dissimilar, equivariance also requires similarity to vary faithfully according to the semantic changes. This allows VLMs to generalize better to nuanced and unseen multimodal compositions. However, modeling equivariance is challenging as the ground truth of semantic change is difficult to collect. For example, given an image-text pair about a dog, it is unclear to what extent the similarity changes when the pixel is changed from dog to cat? To this end, we propose EqSim, a regularization loss that can be efficiently calculated from any two matched training pairs and easily pluggable into existing image-text retrieval fine-tuning. Meanwhile, to further diagnose the equivariance of VLMs, we present a new challenging benchmark EqBen. Compared to the existing evaluation sets, EqBen is the first to focus on "visual-minimal change". Extensive experiments show the lack of equivariance in current VLMs and validate the effectiveness of EqSim.
Graph convolutional networks (GCNs) enable end-to-end learning on graph structured data. However, many works assume a given graph structure. When the input graph is noisy or unavailable, one approach is to construct or learn a latent graph structure. These methods typically fix the choice of node degree for the entire graph, which is suboptimal. Instead, we propose a novel end-to-end differentiable graph generator which builds graph topologies where each node selects both its neighborhood and its size. Our module can be readily integrated into existing pipelines involving graph convolution operations, replacing the predetermined or existing adjacency matrix with one that is learned, and optimized, as part of the general objective. As such it is applicable to any GCN. We integrate our module into trajectory prediction, point cloud classification and node classification pipelines resulting in improved accuracy over other structure-learning methods across a wide range of datasets and GCN backbones.
ACTIVE: Towards Highly Transferable 3D Physical Camouflage for Universal and Robust Vehicle Evasion
PDF ↗Adversarial camouflage has garnered attention for its ability to attack object detectors from any viewpoint by covering the entire object's surface. However, universality and robustness in existing methods often fall short as the transferability aspect is often overlooked, thus restricting their application only to a specific target with limited performance. To address these challenges, we present Adversarial Camouflage for Transferable and Intensive Vehicle Evasion (ACTIVE), a state-of-the-art physical camouflage attack framework designed to generate universal and robust adversarial camouflage capable of concealing any 3D vehicle from detectors. Our framework incorporates innovative techniques to enhance universality and robustness, including a refined texture rendering that enables common texture application to different vehicles without being constrained to a specific texture map, a novel stealth loss that renders the vehicle undetectable, and a smooth and camouflage loss to enhance the naturalness of the adversarial camouflage. Our extensive experiments on 15 different models show that ACTIVE consistently outperforms existing works on various public detectors, including the latest YOLOv7. Notably, our universality evaluations reveal promising transferability to other vehicle classes, tasks (segmentation models), and the real world, not just other vehicles.
Recent studies indicate that Vision Transformers (ViTs) are robust against out-of-distribution scenarios. In particular, the Fully Attentional Network (FAN) - a family of ViT backbones, has achieved state-of-the-art robustness. In this paper, we revisit the FAN models and improve their pre-training with a self-emerging token labeling (STL) framework. Our method contains a two-stage training framework. Specifically, we first train a FAN token labeler (FAN-TL) to generate semantically meaningful patch token labels, followed by a FAN student model training stage that uses both the token labels and the original class label. With the proposed STL framework, our best model based on FAN-L-Hybrid (77.3M parameters) achieves 84.8% Top-1 accuracy and 42.1% mCE on ImageNet-1K and ImageNet-C, and sets a new state-of-the-art for ImageNet-A (46.1%) and ImageNet-R (56.6%) without using extra data, outperforming the original FAN counterpart by significant margins. The proposed framework also demonstrates significantly enhanced performance on downstream tasks such as semantic segmentation, with up to 1.7% improvement in robustness over the counterpart model.
Motivated by concerns that large-scale diffusion models can produce undesirable output such as sexually explicit content or copyrighted artistic styles, we study erasure of specific concepts from diffusion model weights. We propose a fine-tuning method that can erase a visual concept from a pre-trained diffusion model, given only the name of the style and using negative guidance as a teacher. We benchmark our method against previous approaches that remove sexually explicit content and demonstrate its effectiveness, performing on par with Safe Latent Diffusion and censored training. To evaluate artistic style removal, we conduct experiments erasing five modern artists from the network and conduct a user study to assess the human perception of the removed styles. Unlike previous methods, our approach can remove concepts from a diffusion model permanently rather than modifying the output at the inference time, so it cannot be circumvented even if a user has access to model weights. Our code, data, and results are available at erasing.baulab.info
Large-scale data from real-world usually follow a long-tailed distribution (i.e., a few majority classes occupy plentiful training data, while most minority classes have few samples), making the hyperplanes heavily skewed to the minority classes. Traditionally, reweighting is adopted to make the hyperplanes fairly split the feature space, where the weights are designed according to the number of samples. However, we find that the number of samples in a class can not accurately measure the size of its spanned space, especially for the majority class, where the size of its spanned space is usually larger than the samples' number because of the high diversity. Therefore, weights designed based on the samples' number will still compress the space of minority classes. In this paper, we reconsider reweighting from a totally new perspective of analyzing the spanned space of each class. We argue that, besides statistical numbers, relations between samples are also significant for sufficiently depicting the spanned space. Consequently, we estimate the size of the spanned space for each category, namely effective area, by detailedly analyzing its samples' distribution. By treating samples of a class as identically distributed random variables and analyzing their correlations, a simple and non-parametric formula is derived to estimate the effective area. Then, the weight simply calculated inversely proportional to the effective area of each class is adopted to achieve fairer training. Note that our weights are more flexible as they can be adaptively adjusted along with the optimizing features during training. Experiments on four long-tailed datasets show that the proposed weights outperform the state-of-the-art reweighting methods. Moreover, our method can also achieve better results on statistically balanced CIFAR-10/100. Code is available at https://github.com/xiaohua-chen/AREA.
We address outdoor Neural Radiance Fields (NeRF) with LiDAR maps. Existing NeRF methods usually require specially collected hypersampled source views and do not perform well with the open source camera-LiDAR datasets - significantly limiting the approach's practical utility. In this paper, we demonstrate an approach that allows for these datasets to be utilized for high quality neural renderings. Our design leverages 1) LiDAR sensors for strong 3D geometry priors that significantly improve the ray sampling locality, and 2) Conditional Adversarial Networks (cGANs) to recover image details since aggregating embeddings from imperfect LiDAR maps causes artifacts in the synthesized images. Our experiments show that while NeRF baselines produce either noisy or blurry results on Argoverse 2, the images synthesized by our system not only outperform baselines in image quality metrics under both clean and noisy conditions, but also obtain closer Detectron2 results to the ground truth images. Furthermore, to show the substantial applicability of our system, we demonstrate that our system can be used in data augmentation for training a pose regression network and multi-season view synthesis. Our dataset and code will be released
Learning to Generate Semantic Layouts for Higher Text-Image Correspondence in Text-to-Image Synthesis
PDF ↗Existing text-to-image generation approaches have set high standards for photorealism and text-image correspondence, largely benefiting from web-scale text-image datasets, which can include up to 5 billion pairs. However, text-to-image generation models trained on domain-specific datasets, such as urban scenes, medical images, and faces, still suffer from low text-image correspondence due to the lack of text-image pairs. Additionally, collecting billions of text-image pairs for a specific domain can be time-consuming and costly. Thus, ensuring high text-image correspondence without relying on web-scale text-image datasets remains a challenging task. In this paper, we present a novel approach for enhancing text-image correspondence by leveraging available semantic layouts. Specifically, we propose a Gaussian-categorical diffusion process that simultaneously generates both images and corresponding layout pairs. Our experiments reveal that we can guide text-to-image generation models to be aware of the semantics of different image regions, by training the model to generate semantic labels for each pixel. We demonstrate that our approach achieves higher text-image correspondence compared to existing text-to-image generation approaches in the Multi-Modal CelebA-HQ and the Cityscapes dataset, where text-image pairs are scarce.
Self-supervised representation learning follows a paradigm of withholding some part of the data and tasking the network to predict it from the remaining part. Among many techniques, data augmentation lies at the core for creating the information gap. Towards this end, masking has emerged as a generic and powerful tool where content is withheld along the sequential dimension, e.g., spatial in images, temporal in audio, and syntactic in language. In this paper, we explore the orthogonal channel dimension for generic data augmentation by exploiting precision redundancy. The data for each channel is quantized through a non-uniform quantizer, with the quantized value sampled randomly within randomly sampled quantization bins. From another perspective, quantization is analogous to channel-wise masking, as it removes the information within each bin, but preserves the information across bins. Our approach significantly surpasses existing generic data augmentation methods, while showing on par performance against modality- specific augmentations. We comprehensively evaluate our approach on vision, audio, 3D point clouds, as well as the DABS benchmark which is comprised of various data modalities. The code is available at https: //github.com/microsoft/random_quantize.
While 3D body reconstruction methods have made remarkable progress recently, it remains difficult to acquire the sufficiently accurate and numerous 3D supervisions required for training. In this paper, we propose KNOWN, a framework that effectively utilizes body KNOWledge and uNcertainty modeling to compensate for insufficient 3D supervisions. KNOWN exploits a comprehensive set of generic body constraints derived from well-established body knowledge. These generic constraints precisely and explicitly characterize the reconstruction plausibility and enable 3D reconstruction models to be trained without any 3D data. Moreover, existing methods typically use images from multiple datasets during training, which can result in data noise (e.g., inconsistent joint annotation) and data imbalance (e.g., minority images representing unusual poses or captured from challenging camera views). KNOWN solves these problems through a novel probabilistic framework that models both aleatoric and epistemic uncertainty. Aleatoric uncertainty is encoded in a robust Negative Log-Likelihood (NLL) training loss, while epistemic uncertainty is used to guide model refinement. Experiments demonstrate that KNOWN's body reconstruction outperforms prior weakly-supervised approaches, particularly on the challenging minority images.
Network calibration aims to accurately estimate the level of confidences, which is particularly important for employing deep neural networks in real-world systems. Recent approaches leverage mixup to calibrate the network's predictions during training. However, they do not consider the problem that mixtures of labels in mixup may not accurately represent the actual distribution of augmented samples. In this paper, we present RankMixup, a novel mixup-based framework alleviating the problem of the mixture of labels for network calibration. To this end, we propose to use an ordinal ranking relationship between raw and mixup-augmented samples as an alternative supervisory signal to the label mixtures for network calibration. We hypothesize that the network should estimate a higher level of confidence for the raw samples than the augmented ones (Fig.1). To implement this idea, we introduce a mixup-based ranking loss (MRL) that encourages lower confidences for augmented samples compared to raw ones, maintaining the ranking relationship. We also propose to leverage the ranking relationship among multiple mixup-augmented samples to further improve the calibration capability. Augmented samples with larger mixing coefficients are expected to have higher confidences and vice versa (Fig.1). That is, the order of confidences should be aligned with that of mixing coefficients. To this end, we introduce a novel loss, M-NDCG, in order to reduce the number of misaligned pairs of the coefficients and confidences. Extensive experimental results on standard benchmarks for network calibration demonstrate the effectiveness of RankMixup.
StyleDomain: Efficient and Lightweight Parameterizations of StyleGAN for One-shot and Few-shot Domain Adaptation
PDF ↗Domain adaptation of GANs is a problem of fine-tuning GAN models pretrained on a large dataset (e.g. StyleGAN) to a specific domain with few samples (e.g. painting faces, sketches, etc.). While there are many methods that tackle this problem in different ways, there are still many important questions that remain unanswered. In this paper, we provide a systematic and in-depth analysis of the domain adaptation problem of GANs, focusing on the StyleGAN model. We perform a detailed exploration of the most important parts of StyleGAN that are responsible for adapting the generator to a new domain depending on the similarity between the source and target domains. As a result of this study, we propose new efficient and lightweight parameterizations of StyleGAN for domain adaptation. Particularly, we show that there exist directions in StyleSpace (StyleDomain directions) that are sufficient for adapting to similar domains. For dissimilar domains, we propose Affine+ and AffineLight+ parameterizations that allows us to outperform existing baselines in few-shot adaptation while having significantly less training parameters. Finally, we examine StyleDomain directions and discover their many surprising properties that we apply for domain mixing and cross-domain image morphing. Source code can be found at https://github.com/AIRI-Institute/StyleDomain.
Recently, there has been a growing interest in learning-based explicit methods due to their ability to respect the original input and preserve details. However, the connectivity on complex structures is still difficult to infer due to the limited local shape perception, resulting in artifacts and non-watertight triangles. In this paper, we present a novel learning-based method with Delaunay triangulation to achieve high-precision reconstruction. We model the Delaunay triangulation as a dual graph, extract local geometric information from the points, and embed it into the structural representation of Delaunay triangulation in an organic way, benefiting fine-grained details reconstruction. To encourage neighborhood information interaction of edges and nodes in the graph, we introduce a local graph iteration algorithm, which is a variant of graph neural network. Moreover, a geometric constraint loss further improves the classification of tetrahedrons. Benefiting from our fully local network, a scaling strategy is designed to enable large-scale reconstruction. Experiments show that our method yields watertight and high-quality meshes. Especially for some thin structures and sharp edges, our method shows better performance than the current state-of-the-art methods. Furthermore, it has a strong adaptability to point clouds of different densities.
Unaligned 2D to 3D Translation with Conditional Vector-Quantized Code Diffusion using Transformers
PDF ↗Generating 3D images of complex objects conditionally from a few 2D views is a difficult synthesis problem, compounded by issues such as domain gap and geometric misalignment. For instance, a unified framework such as Generative Adversarial Networks cannot achieve this unless they explicitly define both a domain-invariant and geometric-invariant joint latent distribution, whereas Neural Radiance Fields are generally unable to handle both issues as they optimize at the pixel level. By contrast, we propose a simple and novel 2D to 3D synthesis approach based on conditional diffusion with vector-quantized codes. Operating in an information-rich code space enables high-resolution 3D synthesis via full-coverage attention across the views. Specifically, we generate the 3D codes (e.g. for CT images) conditional on previously generated 3D codes and the entire codebook of two 2D views (e.g. 2D X-rays). Qualitative and quantitative results demonstrate state-of-the-art performance over specialized methods across varied evaluation criteria, including fidelity metrics such as density, coverage, and distortion metrics for two complex volumetric imagery datasets from in real-world scenarios.