CVPR 2022 论文列表

Novel object captioning aims at describing objects absent from training data, with the key ingredient being the provision of object vocabulary to the model. Although existing methods heavily rely on an object detection model, we view the detection step as vocabulary retrieval from an external knowledge in the form of embeddings for any object's definition from Wiktionary, where we use in the retrieval image region features learned from a transformers model. We propose an end-to-end Novel Object Captioning with Retrieved vocabulary from External Knowledge method (NOC-REK), which simultaneously learns vocabulary retrieval and caption generation, successfully describing novel objects outside of the training dataset. Furthermore, our model eliminates the requirement for model retraining by simply updating the external knowledge whenever a novel object appears. Our comprehensive experiments on held-out COCO and Nocaps datasets show that our NOC-REK is considerably effective against SOTAs.

Neural image caption generation (NICG) models have received massive attention from the research community due to their excellent performance in visual understanding. Existing work focuses on improving NICG model accuracy while efficiency is less explored. However, many real-world applications require real-time feedback, which highly relies on the efficiency of NICG models. Recent research observed that the efficiency of NICG models could vary for different inputs. This observation brings in a new attack surface of NICG models, i.e., An adversary might be able to slightly change inputs to cause the NICG models to consume more computational resources. To further understand such efficiency-oriented threats, we propose a new attack approach, NICGSlowDown, to evaluate the efficiency robustness of NICG models. Our experimental results show that NICGSlowDown can generate images with human-unnoticeable perturbations that will increase the NICG model latency up to 483.86%. We hope this research could raise the community's concern about the efficiency robustness of NICG models.

Neural implicit representations have recently shown encouraging results in various domains, including promising progress in simultaneous localization and mapping (SLAM). Nevertheless, existing methods produce over-smoothed scene reconstructions and have difficulty scaling up to large scenes. These limitations are mainly due to their simple fully-connected network architecture that does not incorporate local information in the observations. In this paper, we present NICE-SLAM, a dense SLAM system that incorporates multi-level local information by introducing a hierarchical scene representation. Optimizing this representation with pre-trained geometric priors enables detailed reconstruction on large indoor scenes. Compared to recent neural implicit SLAM systems, our approach is more scalable, efficient, and robust. Experiments on five challenging datasets demonstrate competitive results of NICE-SLAM in both mapping and tracking quality. Project page: https://pengsongyou.github.io/nice-slam

Person re-identification aims to retrieve persons in highly varying settings across different cameras and scenarios, in which robust and discriminative representation learning is crucial. Most research considers learning representations from single images, ignoring any potential interactions between them. However, due to the high intra-identity variations, ignoring such interactions typically leads to outlier features. To tackle this issue, we propose a Neighbor Transformer Network, or NFormer, which explicitly models interactions across all input images, thus suppressing outlier features and leading to more robust representations overall. As modelling interactions between enormous amount of images is a massive task with lots of dis- tractors, NFormer introduces two novel modules, the Landmark Agent Attention, and the Reciprocal Neighbor Softmax. Specifically, the Landmark Agent Attention efficiently models the relation map between images by a low-rank factorization with a few landmarks in feature space. Moreover, the Reciprocal Neighbor Softmax achieves sparse attention to relevant -rather than all- neighbors only, which alleviates interference of irrelevant representations and further relieves the computational burden. In experiments on four large-scale datasets, NFormer achieves a new state-of-the-art. The code is released at https://github.com/haochenheheda/NFormer.

Burst denoising is now more relevant than ever, as computational photography helps overcome sensitivity issues inherent in mobile phones and small cameras. A major challenge in burst-denoising is in coping with pixel misalignment, which was so far handled with rather simplistic assumptions of simple motion, or the ability to align in pre-processing. Such assumptions are not realistic in the presence of large motion and high levels of noise. We show that Neural Radiance Fields (NeRFs), originally suggested for physics-based novel-view rendering, can serve as a powerful framework for burst denoising. NeRFs have an inherent capability of handling noise as they integrate information from multiple images, but they are limited in doing so, mainly since they build on pixel-wise operations which are suitable to ideal imaging conditions. Our approach, termed NAN, leverages inter-view and spatial information in NeRFs to better deal with noise. It achieves state-of-the-art results in burst denoising and is especially successful in coping with large movement and occlusions, under very high levels of noise. With the rapid advances in accelerating NeRFs, it could provide a powerful platform for denoising in challenging environments.

Deep cross-modal hashing has become an essential tool for supervised multimodal search. These models tend to be optimized with large, curated multimodal datasets, where most labels have been manually verified. Unfortunately, in many scenarios, such accurate labeling may not be available. In contrast, datasets with low-quality annotations may be acquired, which inevitably introduce numerous mistakes or label noise and therefore degrade the search performance. To address the challenge, we present a general robust cross-modal hashing framework to correlate distinct modalities and combat noisy labels simultaneously. More specifically, we propose a proxy-based contrastive (PC) loss to mitigate the gap between different modalities and train networks for different modalities jointly with small-loss samples that are selected with the PC loss and a mutual quantization loss. The small-loss sample selection from such joint loss can help choose confident examples to guide the model training, and the mutual quantization loss can maximize the agreement between different modalities and is beneficial to improve the effectiveness of sample selection. Experiments on three widely-used multimodal datasets show that our method significantly outperforms existing state-of-the-art methods.

Pan-sharpening aims to integrate the complementary information of texture-rich PAN images and multi-spectral (MS) images to produce the texture-rich MS images. Despite the remarkable progress, existing state-of-the-art Pan-sharpening methods don't explicitly enforce the complementary information learning between two modalities of PAN and MS images. This leads to information redundancy not being handled well, which further limits the performance of these methods. To address the above issue, we propose a novel mutual information-driven Pan-sharpening framework in this paper. To be specific, we first project the PAN and MS image into modality-aware feature space independently, and then impose the mutual information minimization over them to explicitly encourage the complementary information learning. Such operation is capable of reducing the information redundancy and improving the model performance. Extensive experimental results over multiple satellite datasets demonstrate that the proposed algorithm outperforms other state-of-the-art methods qualitatively and quantitatively with great generalization ability to real-world scenes.

Video understanding requires reasoning at multiple spatiotemporal resolutions -- from short fine-grained motions to events taking place over longer durations. Although transformer architectures have recently advanced the state-of-the-art, they have not explicitly modelled different spatiotemporal resolutions. To this end, we present Multiview Transformers for Video Recognition (MTV). Our model consists of separate encoders to represent different views of the input video with lateral connections to fuse information across views. We present thorough ablation studies of our model and show that MTV consistently performs better than single-view counterparts in terms of accuracy and computational cost across a range of model sizes. Furthermore, we achieve state-of-the-art results on six standard datasets, and improve even further with large-scale pretaining. Code and checkpoints are available at: https://github.com/google-research/scenic.

Many adaptations of transformers have emerged to address the single-modal vision tasks, where self-attention modules are stacked to handle input sources like images. Intuitively, feeding multiple modalities of data to vision transformers could improve the performance, yet the inner-modal attentive weights may be diluted, which could thus greatly undermine the final performance. In this paper, we propose a multimodal token fusion method (TokenFusion), tailored for transformer-based vision tasks. To effectively fuse multiple modalities, TokenFusion dynamically detects uninformative tokens and substitute these tokens with projected and aggregated inter-modal features. Residual positional alignment is also adopted to enable explicit utilization of the inter-modal alignments after fusion. The design of TokenFusion allows the transformer to learn correlations among multimodal features, while the single-modal transformer architecture remains largely intact. Extensive experiments are conducted on a variety of homogeneous and heterogeneous modalities and demonstrate that TokenFusion surpasses state-of-the-art methods in three typical vision tasks: multimodal image-to-image translation, RGB-depth semantic segmentation, and 3D object detection with point cloud and images.

Recognition of materials from their visual appearance is essential for computer vision tasks, especially those that involve interaction with the real world. Material segmentation, i.e., dense per-pixel recognition of materials, remains challenging as, unlike objects, materials do not exhibit clearly discernible visual signatures in their regular RGB appearances. Different materials, however, do lead to different radiometric behaviors, which can often be captured with non-RGB imaging modalities. We realize multimodal material segmentation from RGB, polarization, and near-infrared images. We introduce the MCubeS dataset (from MultiModal Material Segmentation) which contains 500 sets of multimodal images capturing 42 street scenes. Ground truth material segmentation as well as semantic segmentation are annotated for every image and pixel. We also derive a novel deep neural network, MCubeSNet, which learns to focus on the most informative combinations of imaging modalities for each material class with a newly derived region-guided filter selection (RGFS) layer. We use semantic segmentation, as a prior to "guide" this filter selection. To the best of our knowledge, our work is the first comprehensive study on truly multimodal material segmentation. We believe our work opens new avenues of practical use of material information in safety critical applications.

Integration of heterogeneous and high-dimensional data (e.g., multiomics) is becoming increasingly important. Existing multimodal classification algorithms mainly focus on improving performance by exploiting the complementarity from different modalities. However, conventional approaches are basically weak in providing trustworthy multimodal fusion, especially for safety-critical applications (e.g., medical diagnosis). For this issue, we propose a novel trustworthy multimodal classification algorithm termed Multimodal Dynamics, which dynamically evaluates both the feature-level and modality-level informativeness for different samples and thus trustworthily integrates multiple modalities. Specifically, a sparse gating is introduced to capture the information variation of each within-modality feature and the true class probability is employed to assess the classification confidence of each modality. Then a transparent fusion algorithm based on the dynamical informativeness estimation strategy is induced. To the best of our knowledge, this is the first work to jointly model both feature and modality variation for different samples to provide trustworthy fusion in multi-modal classification. Extensive experiments are conducted on multimodal medical classification datasets. In these experiments, superior performance and trustworthiness of our algorithm are clearly validated compared to the state-of-the-art methods.

Reconstruction of geometric structures from images using supervised learning suffers from limited available amount of accurate data. One type of such data is accurate real-world RGB-D images. A major challenge in acquiring such ground truth data is the accurate alignment between RGB images and the point cloud measured by a depth scanner. To overcome this difficulty, we consider a differential optimization method that aligns a colored point cloud with a given color image through iterative geometric and color matching. In the proposed framework, the optimization minimizes the photometric difference between the colors of the point cloud and the corresponding colors of the image pixels. Unlike other methods that try to reduce this photometric error, we analyze the computation of the gradient on the image plane and propose a different direct scheme. We assume that the colors produced by the geometric scanner camera and the color camera sensor are different and therefore characterized by different chromatic acquisition properties. Under these multimodal conditions, we find the transformation between the camera image and the point cloud colors. We alternately optimize for aligning the position of the point cloud and matching the different color spaces. The alignments produced by the proposed method are demonstrated on both synthetic data with quantitative evaluation and real scenes with qualitative results.

Optimization-based meta-learning offers a promising direction for few-shot learning that is essential for many real-world computer vision applications. However, learning from few samples introduces uncertainty, and quantifying model confidence for few-shot predictions is essential for many critical domains. Furthermore, few-shot tasks used in meta training are usually sampled randomly from a task distribution for an iterative model update, leading to high labeling costs and computational overhead in meta-training. We propose a novel uncertainty-aware task selection model for label efficient meta-learning. The proposed model formulates a multidimensional belief measure, which can quantify the known uncertainty and lower bound the unknown uncertainty of any given task. Our theoretical result establishes an important relationship between the conflicting belief and the incorrect belief. The theoretical result allows us to estimate the total uncertainty of a task, which provides a principled criterion for task selection. A novel multi-query task formulation is further developed to improve both the computational and labeling efficiency of meta-learning. Experiments conducted over multiple real-world few-shot image classification tasks demonstrate the effectiveness of the proposed model.

The 3D visual grounding task aims to ground a natural language description to the targeted object in a 3D scene, which is usually represented in 3D point clouds. Previous works studied visual grounding under specific views. The vision-language correspondence learned by this way can easily fail once the view changes. In this paper, we propose a Multi-View Transformer (MVT) for 3D visual grounding. We project the 3D scene to a multi-view space, in which the position information of the 3D scene under different views are modeled simultaneously and aggregated together. The multi-view space enables the network to learn a more robust multi-modal representation for 3D visual grounding and eliminates the dependence on specific views. Extensive experiments show that our approach significantly outperforms all state-of-the-art methods. Specifically, on Nr3D and Sr3D datasets, our method outperforms the best competitor by 11.2% and 7.1% and even surpasses recent work with extra 2D assistance by 5.9% and 6.6%. Our code is available at https://github.com/sega-hsj/MVT-3DVG.

We propose an analysis-by-synthesis method for fast multi-view 3D reconstruction of opaque objects with arbitrary materials and illumination. State-of-the-art methods use both neural surface representations and neural rendering. While flexible, neural surface representations are a significant bottleneck in optimization runtime. Instead, we represent surfaces as triangle meshes and build a differentiable rendering pipeline around triangle rasterization and neural shading. The renderer is used in a gradient descent optimization where both a triangle mesh and a neural shader are jointly optimized to reproduce the multi-view images. We evaluate our method on a public 3D reconstruction dataset and show that it can match the reconstruction accuracy of traditional baselines and neural approaches while surpassing them in optimization runtime. Additionally, we investigate the shader and find that it learns an interpretable representation of appearance, enabling applications such as 3D material editing.

3D-aware image synthesis aims to generate images of objects from multiple views by learning a 3D representation. However, one key challenge remains: existing approaches lack geometry constraints, hence usually fail to generate multi-view consistent images. To address this challenge, we propose Multi-View Consistent Generative Adversarial Networks (MVCGAN) for high-quality 3D-aware image synthesis with geometry constraints. By leveraging the underlying 3D geometry information of generated images, i.e., depth and camera transformation matrix, we explicitly establish stereo correspondence between views to perform multi-view joint optimization. In particular, we enforce the photometric consistency between pairs of views and integrate a stereo mixup mechanism into the training process, encouraging the model to reason about the correct 3D shape. Besides, we design a two-stage training strategy with feature-level multi-view joint optimization to improve the image quality. Extensive experiments on three datasets demonstrate that MVCGAN achieves the state-of-the-art performance for 3D-aware image synthesis.

Deep learning-based image salient object detection (SOD) heavily relies on large-scale training data with pixel-wise labeling. High-quality labels involve intensive labor and are expensive to acquire. In this paper, we propose a novel multi-source uncertainty mining method to facilitate unsupervised deep learning from multiple noisy labels generated by traditional handcrafted SOD methods. We design an Uncertainty Mining Network (UMNet) which consists of multiple Merge-and-Split (MS) modules to recursively analyze the commonality and difference among multiple noisy labels and infer pixel-wise uncertainty map for each label. Meanwhile, we model the noisy labels using Gibbs distribution and propose a weighted uncertainty loss to jointly train the UMNet with the SOD network. As a consequence, our UMNet can adaptively select reliable labels for SOD network learning. Extensive experiments on benchmark datasets demonstrate that our method not only outperforms existing unsupervised methods, but also is on par with fully-supervised state-of-the-art models.

Video deblurring has achieved remarkable progress thanks to the success of deep neural networks. Most methods solve for the deblurring end-to-end with limited information propagation from the video sequence. However, different frame regions exhibit different characteristics and should be provided with corresponding relevant information. To achieve fine-grained deblurring, we designed a memory branch to memorize the blurry-sharp feature pairs in the memory bank, thus providing useful information for the blurry query input. To enrich the memory of our memory bank, we further designed a bidirectional recurrency and multi-scale strategy based on the memory bank. Experimental results demonstrate that our model outperforms other state-of-the-art methods while keeping the model complexity and inference time low. The code is available at https://github.com/jibo27/MemDeblur.

Vision Transformers (ViTs) have emerged with superior performance on computer vision tasks compared to convolutional neural network (CNN)-based models. However, ViTs are mainly designed for image classification that generate single-scale low-resolution representations, which makes dense prediction tasks such as semantic segmentation challenging for ViTs. Therefore, we propose HRViT, which enhances ViTs to learn semantically-rich and spatially-precise multi-scale representations by integrating high-resolution multi-branch architectures with ViTs. We balance the model performance and efficiency of HRViT by various branch-block co-optimization techniques. Specifically, we explore heterogeneous branch designs, reduce the redundancy in linear layers, and augment the attention block with enhanced expressiveness. Those approaches enabled \ours to push the Pareto frontier of performance and efficiency on semantic segmentation to a new level, as our evaluation results on ADE20K and Cityscapes show. HRViT achieves 50.20% mIoU on ADE20K and 83.16% mIoU on Cityscapes for semantic segmentation tasks, surpassing state-of-the-art MiT and CSWin backbones with an average of +1.78 mIoU improvement, 28% parameter reduction, and 21% FLOPs reduction, demonstrating the potential of HRViT as a strong vision backbone for semantic segmentation.