CVPR 2022 论文列表

Cereal grains are a vital part of human diets and are important commodities for people's livelihood and international trade. Grain Appearance Inspection (GAI) serves as one of the crucial steps for the determination of grain quality and grain stratification for proper circulation, storage and food processing, etc. GAI is routinely performed manually by qualified inspectors with the aid of some hand tools. Automated GAI has the benefit of greatly assisting inspectors with their jobs but has been limited due to the lack of datasets and clear definitions of the tasks. In this paper we formulate GAI as three ubiquitous computer vision tasks: fine-grained recognition, domain adaptation and out-of-distribution recognition. We present a large-scale and publicly available cereal grains dataset called GrainSpace. Specifically, we construct three types of device prototypes for data acquisition, and a total of 5.25 million images determined by professional inspectors. The grain samples including wheat, maize and rice are collected from five countries and more than 30 regions. We also develop a comprehensive benchmark based on semi-supervised learning and self-supervised learning techniques. To the best of our knowledge, GrainSpace is the first publicly released dataset for cereal grain inspection, https://github.com/hellodfan/GrainSpace.

Although supervised deep stereo matching networks have made impressive achievements, the poor generalization ability caused by the domain gap prevents them from being applied to real-life scenarios. In this paper, we propose to leverage the feature of a model trained on large-scale datasets to deal with the domain shift since it has seen various styles of images. With the cosine similarity based cost volume as a bridge, the feature will be grafted to an ordinary cost aggregation module. Despite the broad-spectrum representation, such a low-level feature contains much general information which is not aimed at stereo matching. To recover more task-specific information, the grafted feature is further input into a shallow network to be transformed before calculating the cost. Extensive experiments show that the model generalization ability can be improved significantly with this broad-spectrum and task-oriented feature. Specifically, based on two well-known architectures PSMNet and GANet, our methods are superior to other robust algorithms when transferring from SceneFlow to KITTI 2015, KITTI 2012, and Middlebury. Code is available at https://github.com/SpadeLiu/Graft-PSMNet.

We present Gradient-SDF, a novel representation for 3D geometry that combines the advantages of implict and explicit representations. By storing at every voxel both the signed distance field as well as its gradient vector field, we enhance the capability of implicit representations with approaches originally formulated for explicit surfaces. As concrete examples, we show that (1) the Gradient-SDF allows us to perform direct SDF tracking from depth images, using efficient storage schemes like hash maps, and that (2) the Gradient-SDF representation enables us to perform photometric bundle adjustment directly in a voxel representation (without transforming into a point cloud or mesh), naturally a fully implicit optimization of geometry and camera poses and easy geometry upsampling. Experimental results confirm that this leads to significantly sharper reconstructions. Since the overall SDF voxel structure is still respected, the proposed Gradient-SDF is equally suited for (GPU) parallelization as related approaches.

In this work we demonstrate the vulnerability of vision transformers (ViTs) to gradient-based inversion attacks. During this attack, the original data batch is reconstructed given model weights and the corresponding gradients. We introduce a method, named GradViT, that optimizes random noise into naturally looking images via an iterative process. The optimization objective consists of (i) a loss on matching the gradients, (ii) image prior in the form of distance to batch normalization statistics of a pretrained CNN model, and (iii) a total variation regularization on patches to guide correct recovery locations. We propose a unique loss scheduling function to overcome local minima during optimization. We evaluate GadViT on ImageNet1K and MS-Celeb-1M datasets, and observe unprecedentedly high fidelity and closeness to the original (hidden) data. During the analysis we find that vision transformers are significantly more vulnerable than previously studied CNNs due to the presence of the attention mechanism. Our method demonstrates new state-of-the-art results for gradient inversion in both qualitative and quantitative metrics. Project page at https://gradvit.github.io.

In 2D-to-3D pose estimation, it is important to exploit the spatial constraints of 2D joints, but it is not yet well modeled. To better model the relation of joints for 3D pose estimation, we propose an effective but simple network, called GraFormer, where a novel transformer architecture is designed via embedding graph convolution layers after multi-head attention block. The proposed GraFormer is built by repeatedly stacking the GraAttention block and the ChebGConv block. The proposed GraAttention block is a new transformer block designed for processing graph-structured data, which is able to learn better features through capturing global information from all the nodes as well as the explicit adjacency structure of nodes. To model the implicit high-order connection relations among non-neighboring nodes, the ChebGConv block is introduced to exchange information between non-neighboring nodes and attain a larger receptive field. We have empirically shown the superiority of GraFormer through extensive experiments on popular public datasets. Specifically, GraFormer outperforms the state-of-the-art GraghSH on the Human3.6M dataset yet only contains 18% parameters of it

We propose a novel method of registering less-overlap RGB-D scans. Our method learns global information of a scene to construct a panorama, and aligns RGB-D scans to the panorama to perform registration. Different from existing methods that use local feature points to register less-overlap RGB-D scans and mismatch too much, we use global information to guide the registration, thereby alleviating the mismatching problem by preserving global consistency of alignments. To this end, we build a scene inference network to construct the panorama representing global information. We introduce a reinforcement learning strategy to iteratively align RGB-D scans with the panorama and refine the panorama representation, which reduces the noise of global information and preserves global consistency of both geometric and photometric alignments. Experimental results on benchmark datasets including SUNCG, Matterport, and ScanNet show the superiority of our method.

The crux of long-term tracking lies in the difficulty of tracking the target with discontinuous moving caused by out-of-view or occlusion. Existing long-term tracking methods follow two typical strategies. The first strategy employs a local tracker to perform smooth tracking and uses another re-detector to detect the target when the target is lost. While it can exploit the temporal context like historical appearances and locations of the target, a potential limitation of such strategy is that the local tracker tends to misidentify a nearby distractor as the target instead of activating the re-detector when the real target is out of view. The other long-term tracking strategy tracks the target in the entire image globally instead of local tracking based on the previous tracking results. Unfortunately, such global tracking strategy cannot leverage the temporal context effectively. In this work, we combine the advantages of both strategies: tracking the target in a global view while exploiting the temporal context. Specifically, we perform global tracking via ensemble of local trackers spreading the full image. The smooth moving of the target can be handled steadily by one local tracker. When the local tracker accidentally loses the target due to suddenly discontinuous moving, another local tracker close to the target is then activated and can readily take over the tracking to locate the target. While the activated local tracker performs tracking locally by leveraging the temporal context, the ensemble of local trackers renders our model the global view for tracking. Extensive experiments on six datasets demonstrate that our method performs favorably against state-of-the-art algorithms.

We present a novel transformer-based architecture for global multi-object tracking. Our network takes a short sequence of frames as input and produces global trajectories for all objects. The core component is a global tracking transformer that operates on objects from all frames in the sequence. The transformer encodes object features from all frames, and uses trajectory queries to group them into trajectories. The trajectory queries are object features from a single frame and naturally produce unique trajectories. Our global tracking transformer does not require intermediate pairwise grouping or combinatorial association, and can be jointly trained with an object detector. It achieves competitive performance on the popular MOT17 benchmark, with 75.3 MOTA and 59.1 HOTA. More importantly, our framework seamlessly integrates into state-of-the-art large-vocabulary detectors to track any objects. Experiments on the challenging TAO dataset show that our framework consistently improves upon baselines that are based on pairwise association, outperforming published work by a significant 7.7 tracking mAP. Code is available at https://github.com/xingyizhou/GTR.

Recently, deep network-based image compressed sensing methods achieved high reconstruction quality and reduced computational overhead compared with traditional methods. However, existing methods obtain measurements only from partial features in the network and use it only once for image reconstruction. They ignore there are low, mid, and high-level features in the network and all of them are essential for high-quality reconstruction. Moreover, using measurements only once may not be enough for extracting richer information from measurements. To address these issues, we propose a novel Measurements Reuse Convolutional Compressed Sensing Network (MR-CCSNet) which employs Global Sensing Module (GSM) to collect all level features for achieving an efficient sensing and Measurements Reuse Block (MRB) to reuse measurements multiple times on multi-scale. Finally, we conduct a series of experiments on three benchmark datasets to show that our model can significantly outperform state-of-the-art methods. Code is available at https://github.com/fze0012/MR-CCSNet.

Optical flow estimation is a fundamental task in computer vision. Recent direct-regression methods using deep neural networks achieve remarkable performance improvement. However, they do not explicitly capture long-term motion correspondences and thus cannot handle large motions effectively. In this paper, inspired by the traditional matching-optimization methods where matching is introduced to handle large displacements before energy-based optimizations, we introduce a simple but effective global matching step before the direct regression and develop a learning-based matching-optimization framework, namely GMFlowNet. In GMFlowNet, global matching is efficiently calculated by applying argmax on 4D cost volumes. Additionally, to improve the matching quality, we propose patch-based overlapping attention to extract large context features. Extensive experiments demonstrate that GMFlowNet outperforms RAFT, the most popular optimization-only method, by a large margin and achieves state-of-the-art performance on standard benchmarks. Thanks to the matching and overlapping attention, GMFlowNet obtains major improvements on the predictions for textureless regions and large motions. Our code is made publicly available at https://github.com/xiaofeng94/GMFlowNet.

Model-agnostic meta-learning (MAML) and its variants have become popular approaches for few-shot learning. However, due to the non-convexity of deep neural nets (DNNs) and the bi-level formulation of MAML, the theoretical properties of MAML with DNNs remain largely unknown. In this paper, we first prove that MAML with over-parameterized DNNs is guaranteed to converge to global optima at a linear rate. Our convergence analysis indicates that MAML with over-parameterized DNNs is equivalent to kernel regression with a novel class of kernels, which we name as Meta Neural Tangent Kernels (MetaNTK). Then, we propose MetaNTK-NAS, a new training-free neural architecture search (NAS) method for few-shot learning that uses MetaNTK to rank and select architectures. Empirically, we compare our MetaNTK-NAS with previous NAS methods on two popular few-shot learning benchmarks, miniImageNet, and tieredImageNet. We show that the performance of MetaNTK-NAS is comparable or better than the state-of-the-art NAS method designed for few-shot learning while enjoying more than 100x speedup. We believe the efficiency of MetaNTK-NAS makes itself more practical for many real-world tasks.

The integration of Vector Quantised Variational AutoEncoder (VQ-VAE) with autoregressive models as generation part has yielded high-quality results on image generation. However, the autoregressive models will strictly follow the progressive scanning order during the sampling phase. This leads the existing VQ series models to hardly escape the trap of lacking global information. Denoising Diffusion Probabilistic Models (DDPM) in the continuous domain have shown a capability to capture the global context, while generating high-quality images. In the discrete state space, some works have demonstrated the potential to perform text generation and low resolution image generation. We show that with the help of a content-rich discrete visual codebook from VQ-VAE, the discrete diffusion model can also generate high fidelity images with global context, which compensates for the deficiency of the classical autoregressive model along pixel space. Meanwhile, the integration of the discrete VAE with the diffusion model resolves the drawback of conventional autoregressive models being oversized, and the diffusion model which demands excessive time in the sampling process when generating images. It is found that the quality of the generated images is heavily dependent on the discrete visual codebook. Extensive experiments demonstrate that the proposed Vector Quantised Discrete Diffusion Model (VQ-DDM) is able to achieve comparable performance to top-tier methods with low complexity. It also demonstrates outstanding advantages over other vectors quantised with autoregressive models in terms of image inpainting tasks without additional training.

Attaching attributes (such as color, shape, state, action) to object categories is an important computer vision problem. Attribute prediction has seen exciting recent progress and is often formulated as a multi-label classification problem. Yet significant challenges remain in: 1) predicting a large number of attributes over multiple object categories, 2) modeling category-dependence of attributes, 3) methodically capturing both global and local scene context, and 4) robustly predicting attributes of objects with low pixel-count. To address these issues, we propose a novel multi-category attribute prediction deep architecture named GlideNet, which contains three distinct feature extractors. A global feature extractor recognizes what objects are present in a scene, whereas a local one focuses on the area surrounding the object of interest. Meanwhile, an intrinsic feature extractor uses an extension of standard convolution dubbed Informed Convolution to retrieve features of objects with low pixel-count utilizing its binary mask. GlideNet then uses gating mechanisms with binary masks and its self-learned category embedding to combine the dense embeddings. Collectively, the Global-Local-Intrinsic blocks comprehend the scene's global context while attending to the characteristics of the local object of interest. The architecture adapts the feature composition based on the category via category embedding. Finally, using the combined features, an interpreter predicts the attributes, and the length of the output is determined by the category, thereby removing unnecessary attributes. GlideNet can achieve compelling results on two recent and challenging datasets -- VAW and CAR -- for large-scale attribute prediction. For instance, it obtains more than 5% gain over state of the art in the mean recall (mR) metric. GlideNet's advantages are especially apparent when predicting attributes of objects with low pixel counts as well as attributes that demand global context understanding. Finally, we show that GlideNet excels in training starved real-world scenarios.

We investigate the problem of training generative models on very sparse collections of 3D models. Particularly, instead of using difficult-to-obtain large sets of 3D models, we demonstrate that geometrically-motivated energy functions can be used to effectively augment and boost only a sparse collection of example (training) models. Technically, we analyze the Hessian of the as-rigid-as-possible (ARAP) energy to adaptively sample from and project to the underlying (local) shape space, and use the augmented dataset to train a variational autoencoder (VAE). We iterate the process, of building latent spaces of VAE and augmenting the associated dataset, to progressively reveal a richer and more expressive generative space for creating geometrically and semantically valid samples. We evaluate our method against a set of strong baselines, provide ablation studies, and demonstrate application towards establishing shape correspondences. GLASS produces multiple interesting and meaningful shape variations even when starting from as few as 3-10 training shapes. Our code is available at https: //sanjeevmk.github.io/glass_webpage/.

Transparent and semi-transparent materials pose significant challenges for existing scene understanding and segmentation algorithms due to their lack of RGB texture which impedes the extraction of meaningful features. In this work, we exploit that the light-matter interactions on glass materials provide unique intensity-polarization cues for each observed wavelength of light. We present a novel learning-based glass segmentation network that leverages both trichromatic (RGB) intensities as well as trichromatic linear polarization cues from a single photograph captured without making any assumption on the polarization state of the illumination. Our novel network architecture dynamically fuses and weights both the trichromatic color and polarization cues using a novel global-guidance and multi-scale self-attention module, and leverages global cross-domain contextual information to achieve robust segmentation. We train and extensively validate our segmentation method on a new large-scale RGB-Polarization dataset (RGBP-Glass), and demonstrate that our method outperforms state-of-the-art segmentation approaches by a significant margin.

Neural architectures based on attention such as vision transformers are revolutionizing image recognition. Their main benefit is that attention allows reasoning about all parts of a scene jointly. In this paper, we show how the global reasoning of (scaled) dot-product attention can be the source of a major vulnerability when confronted with adversarial patch attacks. We provide a theoretical understanding of this vulnerability and relate it to an adversary's ability to misdirect the attention of all queries to a single key token under the control of the adversarial patch. We propose novel adversarial objectives for crafting adversarial patches which target this vulnerability explicitly. We show the effectiveness of the proposed patch attacks on popular image classification (ViTs and DeiTs) and object detection models (DETR). We find that adversarial patches occupying 0.5% of the input can lead to robust accuracies as low as 0% for ViT on ImageNet, and reduce the mAP of DETR on MS COCO to less than 3%.

Despite the substantial progress of deep models for crack recognition, due to the inconsistent cracks in varying sizes, shapes, and noisy background textures, there still lacks the discriminative power of the deeply learned features when supervised by the cross-entropy loss. In this paper, we propose the geometry-aware guided loss (GAGL) that enhances the discrimination ability and is only applied in the training stage without extra computation and memory during inference. The GAGL consists of the feature-based geometry-aware projected gradient descent method (FGA-PGD) that approximates the geometric distances of the features to the class boundaries, and the geometry-aware update rule that learns an anchor of each class as the approximation of the feature expected to have the largest geometric distance to the corresponding class boundary. Then the discriminative power can be enhanced by minimizing the distances between the features and their corresponding class anchors in the feature space. To address the limited availability of related benchmarks, we collect a fully annotated dataset, namely, NPP2021, which involves inconsistent cracks and noisy backgrounds in real-world nuclear power plants. Our proposed GAGL outperforms the state of the arts on various benchmark datasets including CRACK2019, SDNET2018, and our NPP2021.

Research has shown that convolutional neural networks for object recognition are vulnerable to changes in depiction because learning is biased towards the low-level statistics of texture patches. Recent works concentrate on improving robustness by applying style transfer to training examples to mitigate against over-fitting to one depiction style. These new approaches improve performance, but they ignore the geometric variations in object shape that real art exhibits: artists deform and warp objects for artistic effect. Motivated by this observation, we propose a method to reduce bias by jointly increasing the texture and geometry diversities of the training data. In effect, we extend the visual object class to include examples with shape changes that artists use. Specifically, we learn the distribution of warps that cover each given object class. Together with augmenting textures based on a broad distribution of styles, we show by experiments that our method improves performance on several cross-domain benchmarks.

Preserving geometric structures in the scene plays a vital role in image stitching. However, most of the existing methods ignore the large-scale layouts reflected by straight lines or curves, decreasing overall stitching quality. To address this issue, this work presents a structure-preserving stitching approach that produces images with natural visual effects and less distortion. Our method first employs deep learning-based edge detection to extract various types of large-scale edges. Then, the extracted edges are sampled to construct multiple groups of triangles to represent geometric structures. Meanwhile, a GEometric Structure preserving (GES) energy term is introduced to make these triangles undergo similarity transformation. Further, an optimized GES energy term is presented to reasonably determine the weights of the sampling points on the geometric structure, and the term is added into the Global Similarity Prior (GSP) stitching model called GES-GSP to achieve a smooth transition between local alignment and geometric structure preservation. The effectiveness of GES-GSP is validated through comprehensive experiments on a stitching dataset. The experimental results show that the proposed method outperforms several state-of-the-art methods in geometric structure preservation and obtains more natural stitching results. The code and dataset are available at https://github.com/flowerDuo/GES-GSP-Stitching.