我们介绍了光流变压器，被称为流动型，这是一种基于变压器的神经网络体系结构，用于学习光流。流动形式将图像对构建的4D成本量构成，将成本令牌编码为成本记忆，并在新颖的潜在空间中使用备用组变压器（AGT）层编码成本记忆，并通过反复的变压器解码器与动态位置成本查询来解码成本记忆。在SINTEL基准测试中，流动型在干净和最终通行证上达到1.144和2.183平均末端PONIT-ERROR（AEPE），从最佳发布的结果（1.388和2.47）降低了17.6％和11.6％的误差。此外，流程度还达到了强大的概括性能。在不接受Sintel的培训的情况下，FlowFormer在Sintel训练套装清洁通行证上达到了0.95 AEPE，优于最佳发布结果（1.29），提高了26.9％。

基于学习的光流量估计已经与成本量的管道管道，具有用于流回归的卷曲，其固有地限于本地相关性，因此很难解决大型位移的长期挑战。为了缓解这一点，通过大量迭代细化产生一系列流动更新，实现最先进的方法，即筏，逐渐提高其预测的质量，实现了显着的性能，但减慢推理速度。为了实现高精度和效率的光学流量估计，我们通过将光学流作为全球匹配问题重新重新重新重新匹配，完全改造主导流回归管道。具体而言，我们提出了一个GMFlow框架，它由三个主要组件组成：用于功能增强的自定义变压器，全局特征匹配的相关和软邮件，以及用于流传播的自我注意层。此外，我们进一步介绍了一种改进步骤，该步骤在较高分辨率下重复使用GMFlow以进行残余流量预测。我们的新框架优于32次迭代RAFT在挑战的Sintel基准测试中的性能，同时仅使用一个细化并更快地运行，为高效和准确的光学流量估算提供了新的可能性。代码将在https://github.com/haofeixu/gmflow上使用。

We present a unified formulation and model for three motion and 3D perception tasks: optical flow, rectified stereo matching and unrectified stereo depth estimation from posed images. Unlike previous specialized architectures for each specific task, we formulate all three tasks as a unified dense correspondence matching problem, which can be solved with a single model by directly comparing feature similarities. Such a formulation calls for discriminative feature representations, which we achieve using a Transformer, in particular the cross-attention mechanism. We demonstrate that cross-attention enables integration of knowledge from another image via cross-view interactions, which greatly improves the quality of the extracted features. Our unified model naturally enables cross-task transfer since the model architecture and parameters are shared across tasks. We outperform RAFT with our unified model on the challenging Sintel dataset, and our final model that uses a few additional task-specific refinement steps outperforms or compares favorably to recent state-of-the-art methods on 10 popular flow, stereo and depth datasets, while being simpler and more efficient in terms of model design and inference speed.

Optical flow, which computes the apparent motion from a pair of video frames, is a critical tool for scene motion estimation. Correlation volume is the central component of optical flow computational neural models. It estimates the pairwise matching costs between cross-frame features, and is then used to decode optical flow. However, traditional correlation volume is frequently noisy, outlier-prone, and sensitive to motion blur. We observe that, although the recent RAFT algorithm also adopts the traditional correlation volume, its additional context encoder provides semantically representative features to the flow decoder, implicitly compensating for the deficiency of the correlation volume. However, the benefits of this context encoder has been barely discussed or exploited. In this paper, we first investigate the functionality of RAFT's context encoder, then propose a new Context Guided Correlation Volume (CGCV) via gating and lifting schemes. CGCV can be universally integrated with RAFT-based flow computation methods for enhanced performance, especially effective in the presence of motion blur, de-focus blur and atmospheric effects. By incorporating the proposed CGCV with previous Global Motion Aggregation (GMA) method, at a minor cost of 0.5% extra parameters, the rank of GMA is lifted by 23 places on KITTI 2015 Leader Board, and 3 places on Sintel Leader Board. Moreover, at a similar model size, our correlation volume achieves competitive or superior performance to state of the art peer supervised models that employ Transformers or Graph Reasoning, as verified by extensive experiments.

We introduce Recurrent All-Pairs Field Transforms (RAFT), a new deep network architecture for optical flow. RAFT extracts perpixel features, builds multi-scale 4D correlation volumes for all pairs of pixels, and iteratively updates a flow field through a recurrent unit that performs lookups on the correlation volumes. RAFT achieves stateof-the-art performance. On KITTI, RAFT achieves an F1-all error of 5.10%, a 16% error reduction from the best published result (6.10%). On Sintel (final pass), RAFT obtains an end-point-error of 2.855 pixels, a 30% error reduction from the best published result (4.098 pixels). In addition, RAFT has strong cross-dataset generalization as well as high efficiency in inference time, training speed, and parameter count. Code is available at https://github.com/princeton-vl/RAFT.

Optical flow estimation is a classical yet challenging task in computer vision. One of the essential factors in accurately predicting optical flow is to alleviate occlusions between frames. However, it is still a thorny problem for current top-performing optical flow estimation methods due to insufficient local evidence to model occluded areas. In this paper, we propose the Super Kernel Flow Network (SKFlow), a CNN architecture to ameliorate the impacts of occlusions on optical flow estimation. SKFlow benefits from the super kernels which bring enlarged receptive fields to complement the absent matching information and recover the occluded motions. We present efficient super kernel designs by utilizing conical connections and hybrid depth-wise convolutions. Extensive experiments demonstrate the effectiveness of SKFlow on multiple benchmarks, especially in the occluded areas. Without pre-trained backbones on ImageNet and with a modest increase in computation, SKFlow achieves compelling performance and ranks $\textbf{1st}$ among currently published methods on the Sintel benchmark. On the challenging Sintel clean and final passes (test), SKFlow surpasses the best-published result in the unmatched areas ($7.96$ and $12.50$) by $9.09\%$ and $7.92\%$. The code is available at \href{https://github.com/littlespray/SKFlow}{https://github.com/littlespray/SKFlow}.

我们提出了Tain（视频插值的变压器和注意力），这是一个用于视频插值的残留神经网络，旨在插入中间框架，并在其周围连续两个图像框架下进行插值。我们首先提出一个新型的视觉变压器模块，称为交叉相似性（CS），以与预测插值框架相似的外观相似的外观。然后，这些CS特征用于完善插值预测。为了说明CS功能中的遮挡，我们提出了一个图像注意（IA）模块，以使网络可以从另一个框架上关注CS功能。此外，我们还使用封闭式贴片来增强培训数据集，该补丁可以跨帧移动，以改善网络对遮挡和大型运动的稳健性。由于现有方法产生平滑的预测，尤其是在MB附近，因此我们根据图像梯度使用额外的训练损失来产生更清晰的预测。胜过不需要流量估计并与基于流程的方法相当执行的现有方法，同时在VIMEO90K，UCF101和SNU-FILM基准的推理时间上具有计算有效的效率。

我们提出了一个新颖的建筑，以实现密集的对应关系。当前的最新方法是基于变压器的方法，它们专注于功能描述符或成本量集合。但是，尽管关节聚集会通过提供一个人（即图像的结构或语义信息）或像素匹配的相似性来提高一个或另一个，但并非两者都聚集，但并非两者都汇总，尽管关节聚集会相互促进。在这项工作中，我们提出了一个基于变压器的新型网络，该网络以利用其互补信息的方式交织了两种形式的聚合。具体而言，我们设计了一个自我发项层，该层利用描述符来消除嘈杂的成本量，并且还利用成本量以促进准确匹配的方式汇总特征。随后的跨意思层执行进一步的聚合，该聚集在图像的描述上，并由早期层的聚合输出有助于。我们通过层次处理进一步提高了性能，在该处理中，更粗糙的聚合指导那些处于优质水平的过程。我们评估了所提出的方法对密集匹配任务的有效性，并在所有主要基准上实现最先进的性能。还提供了广泛的消融研究来验证我们的设计选择。

We present a compact but effective CNN model for optical flow, called PWC-Net. PWC-Net has been designed according to simple and well-established principles: pyramidal processing, warping, and the use of a cost volume. Cast in a learnable feature pyramid, PWC-Net uses the current optical flow estimate to warp the CNN features of the second image. It then uses the warped features and features of the first image to construct a cost volume, which is processed by a CNN to estimate the optical flow. PWC-Net is 17 times smaller in size and easier to train than the recent FlowNet2 model. Moreover, it outperforms all published optical flow methods on the MPI Sintel final pass and KITTI 2015 benchmarks, running at about 35 fps on Sintel resolution (1024×436) images. Our models are available on https://github.com/NVlabs/PWC-Net.

Astounding results from Transformer models on natural language tasks have intrigued the vision community to study their application to computer vision problems. Among their salient benefits, Transformers enable modeling long dependencies between input sequence elements and support parallel processing of sequence as compared to recurrent networks e.g., Long short-term memory (LSTM). Different from convolutional networks, Transformers require minimal inductive biases for their design and are naturally suited as set-functions. Furthermore, the straightforward design of Transformers allows processing multiple modalities (e.g., images, videos, text and speech) using similar processing blocks and demonstrates excellent scalability to very large capacity networks and huge datasets. These strengths have led to exciting progress on a number of vision tasks using Transformer networks. This survey aims to provide a comprehensive overview of the Transformer models in the computer vision discipline. We start with an introduction to fundamental concepts behind the success of Transformers i.e., self-attention, large-scale pre-training, and bidirectional feature encoding. We then cover extensive applications of transformers in vision including popular recognition tasks (e.g., image classification, object detection, action recognition, and segmentation), generative modeling, multi-modal tasks (e.g., visual-question answering, visual reasoning, and visual grounding), video processing (e.g., activity recognition, video forecasting), low-level vision (e.g., image super-resolution, image enhancement, and colorization) and 3D analysis (e.g., point cloud classification and segmentation). We compare the respective advantages and limitations of popular techniques both in terms of architectural design and their experimental value. Finally, we provide an analysis on open research directions and possible future works. We hope this effort will ignite further interest in the community to solve current challenges towards the application of transformer models in computer vision.

变压器一直是自然语言处理（NLP）和计算机视觉（CV）革命的核心。 NLP和CV的显着成功启发了探索变压器在点云处理中的使用。但是，变压器如何应对点云的不规则性和无序性质？变压器对于不同的3D表示（例如，基于点或体素）的合适性如何？各种3D处理任务的变压器有多大的能力？截至目前，仍然没有对这些问题的研究进行系统的调查。我们第一次为3D点云分析提供了越来越受欢迎的变压器的全面概述。我们首先介绍变压器体系结构的理论，并在2D/3D字段中审查其应用程序。然后，我们提出三种不同的分类法（即实现 - 数据表示和基于任务），它们可以从多个角度对当前的基于变压器的方法进行分类。此外，我们介绍了研究3D中自我注意机制的变异和改进的结果。为了证明变压器在点云分析中的优势，我们提供了基于各种变压器的分类，分割和对象检测方法的全面比较。最后，我们建议三个潜在的研究方向，为3D变压器的开发提供福利参考。

Detection Transformer (DETR) and Deformable DETR have been proposed to eliminate the need for many hand-designed components in object detection while demonstrating good performance as previous complex hand-crafted detectors. However, their performance on Video Object Detection (VOD) has not been well explored. In this paper, we present TransVOD, the first end-to-end video object detection system based on spatial-temporal Transformer architectures. The first goal of this paper is to streamline the pipeline of VOD, effectively removing the need for many hand-crafted components for feature aggregation, e.g., optical flow model, relation networks. Besides, benefited from the object query design in DETR, our method does not need complicated post-processing methods such as Seq-NMS. In particular, we present a temporal Transformer to aggregate both the spatial object queries and the feature memories of each frame. Our temporal transformer consists of two components: Temporal Query Encoder (TQE) to fuse object queries, and Temporal Deformable Transformer Decoder (TDTD) to obtain current frame detection results. These designs boost the strong baseline deformable DETR by a significant margin (2 %-4 % mAP) on the ImageNet VID dataset. TransVOD yields comparable performances on the benchmark of ImageNet VID. Then, we present two improved versions of TransVOD including TransVOD++ and TransVOD Lite. The former fuses object-level information into object query via dynamic convolution while the latter models the entire video clips as the output to speed up the inference time. We give detailed analysis of all three models in the experiment part. In particular, our proposed TransVOD++ sets a new state-of-the-art record in terms of accuracy on ImageNet VID with 90.0 % mAP. Our proposed TransVOD Lite also achieves the best speed and accuracy trade-off with 83.7 % mAP while running at around 30 FPS on a single V100 GPU device. Code and models will be available for further research.

变压器在自然语言处理中的成功最近引起了计算机视觉领域的关注。由于能够学习长期依赖性，变压器已被用作广泛使用的卷积运算符的替代品。事实证明，这种替代者在许多任务中都取得了成功，其中几种最先进的方法依靠变压器来更好地学习。在计算机视觉中，3D字段还见证了使用变压器来增加3D卷积神经网络和多层感知器网络的增加。尽管许多调查都集中在视力中的变压器上，但由于与2D视觉相比，由于数据表示和处理的差异，3D视觉需要特别注意。在这项工作中，我们介绍了针对不同3D视觉任务的100多种变压器方法的系统和彻底审查，包括分类，细分，检测，完成，姿势估计等。我们在3D Vision中讨论了变形金刚的设计，该设计使其可以使用各种3D表示形式处理数据。对于每个应用程序，我们强调了基于变压器的方法的关键属性和贡献。为了评估这些方法的竞争力，我们将它们的性能与12个3D基准测试的常见非转化方法进行了比较。我们通过讨论3D视觉中变压器的不同开放方向和挑战来结束调查。除了提出的论文外，我们的目标是频繁更新最新的相关论文及其相应的实现：https：//github.com/lahoud/3d-vision-transformers。

Convolutional neural networks (CNNs) have recently been very successful in a variety of computer vision tasks, especially on those linked to recognition. Optical flow estimation has not been among the tasks where CNNs were successful. In this paper we construct appropriate CNNs which are capable of solving the optical flow estimation problem as a supervised learning task. We propose and compare two architectures: a generic architecture and another one including a layer that correlates feature vectors at different image locations.Since existing ground truth datasets are not sufficiently large to train a CNN, we generate a synthetic Flying Chairs dataset. We show that networks trained on this unrealistic data still generalize very well to existing datasets such as Sintel and KITTI, achieving competitive accuracy at frame rates of 5 to 10 fps.

视觉变压器正在成为解决计算机视觉问题的强大工具。最近的技术还证明了超出图像域之外的变压器来解决许多与视频相关的任务的功效。其中，由于其广泛的应用，人类的行动识别是从研究界受到特别关注。本文提供了对动作识别的视觉变压器技术的首次全面调查。我们朝着这个方向分析并总结了现有文献和新兴文献，同时突出了适应变形金刚以进行动作识别的流行趋势。由于其专业应用，我们将这些方法统称为``动作变压器''。我们的文献综述根据其架构，方式和预期目标为动作变压器提供了适当的分类法。在动作变压器的背景下，我们探讨了编码时空数据，降低维度降低，框架贴片和时空立方体构造以及各种表示方法的技术。我们还研究了变压器层中时空注意的优化，以处理更长的序列，通常通过减少单个注意操作中的令牌数量。此外，我们还研究了不同的网络学习策略，例如自我监督和零局学习，以及它们对基于变压器的行动识别的相关损失。这项调查还总结了在具有动作变压器重要基准的评估度量评分方面取得的进步。最后，它提供了有关该研究方向的挑战，前景和未来途径的讨论。

光学流量估计是视频分析领域的一个重要而有挑战性问题。卷积神经网络的不同语义级别/层的特征可以提供不同粒度的信息。为了利用如此灵活和全面的信息，我们提出了一个半监督的特征金字塔形相关和残余重建网络（FPCR-Net），用于框架对的光学流量估计。它由两个主要模块组成：金字塔相关映射和剩余重建。金字塔相关映射模块利用全局/本地补丁的多尺度相关性来通过聚合不同尺度的特征来形成多级成本卷。剩余重建模块旨在重建每个阶段中更精细的光学流的子带高频残差。基于金字塔相关映射，我们进一步提出了相关 - 扭曲 - 归一化（CWN）模块，以有效地利用相关性依赖性。实验结果表明，该方案在针对竞争基线方法的平均终点误差（AEE）方面，实现了最先进的性能，改善了0.80,1.15和0.10 - Flownet2，LiteFlowNet和PWC-Net Sintel DataSet的最终通过。

变压器是一种基于关注的编码器解码器架构，彻底改变了自然语言处理领域。灵感来自这一重大成就，最近在将变形式架构调整到计算机视觉（CV）领域的一些开创性作品，这已经证明了他们对各种简历任务的有效性。依靠竞争力的建模能力，与现代卷积神经网络相比在本文中，我们已经为三百不同的视觉变压器进行了全面的审查，用于三个基本的CV任务（分类，检测和分割），提出了根据其动机，结构和使用情况组织这些方法的分类。 。由于培训设置和面向任务的差异，我们还在不同的配置上进行了评估了这些方法，以便于易于和直观的比较而不是各种基准。此外，我们已经揭示了一系列必不可少的，但可能使变压器能够从众多架构中脱颖而出，例如松弛的高级语义嵌入，以弥合视觉和顺序变压器之间的差距。最后，提出了三个未来的未来研究方向进行进一步投资。

动作检测的任务旨在在每个动作实例中同时推论动作类别和终点的本地化。尽管Vision Transformers推动了视频理解的最新进展，但由于在长时间的视频剪辑中，设计有效的架构以进行动作检测是不平凡的。为此，我们提出了一个有效的层次时空时空金字塔变压器（STPT）进行动作检测，这是基于以下事实：变压器中早期的自我注意力层仍然集中在局部模式上。具体而言，我们建议在早期阶段使用本地窗口注意来编码丰富的局部时空时空表示，同时应用全局注意模块以捕获后期的长期时空依赖性。通过这种方式，我们的STPT可以用冗余的大大减少来编码区域和依赖性，从而在准确性和效率之间进行有希望的权衡。例如，仅使用RGB输入，提议的STPT在Thumos14上获得了53.6％的地图，超过10％的I3D+AFSD RGB模型超过10％，并且对使用其他流量的额外流动功能的表现较少，该流量具有31％的GFLOPS ，它是一个有效，有效的端到端变压器框架，用于操作检测。

我们提出了可推广的NERF变压器（GNT），这是一种纯粹的，统一的基于变压器的体系结构，可以从源视图中有效地重建神经辐射场（NERF）。与NERF上的先前作品不同，通过颠倒手工渲染方程来优化人均隐式表示，GNT通过封装两个基于变压器的阶段来实现可概括的神经场景表示和渲染。 GNT的第一阶段，称为View Transformer，利用多视图几何形状作为基于注意力的场景表示的电感偏差，并通过在相邻视图上从异性线中汇总信息来预测与坐标对齐的特征。 GNT的第二阶段，名为Ray Transformer，通过Ray Marching呈现新视图，并使用注意机制直接解码采样点特征的序列。我们的实验表明，当在单个场景上进行优化时，GNT可以在不明确渲染公式的情况下成功重建NERF，甚至由于可学习的射线渲染器，在复杂的场景上甚至将PSNR提高了〜1.3db。当在各种场景中接受培训时，GNT转移到前面的LLFF数据集（LPIPS〜20％，SSIM〜25％$）和合成搅拌器数据集（LPIPS〜20％，SSIM 〜25％$）时，GNN会始终达到最先进的性能4％）。此外，我们表明可以从学习的注意图中推断出深度和遮挡，这意味着纯粹的注意机制能够学习一个物理地面渲染过程。所有这些结果使我们更接近将变形金刚作为“通用建模工具”甚至用于图形的诱人希望。请参阅我们的项目页面以获取视频结果：https：//vita-group.github.io/gnt/。

现有的光流估计器通常采用通常用于图像分类的网络体系结构作为提取人均功能的编码器。但是，由于任务之间的自然差异，用于图像分类的架构可能是最佳的流量估计。为了解决此问题，我们建议一种名为Falownas的神经体系结构搜索方法，以自动找到用于流估计任务的更好的编码器体系结构。我们首先设计一个合适的搜索空间，包括各种卷积运算符，并构建一个体重共享的超级网络，以有效评估候选体系结构。然后，为了更好地训练超级网络，我们提出了特征对齐蒸馏，该蒸馏利用训练有素的流量估计器来指导超级网络的训练。最后，利用资源约束的进化算法找到最佳体系结构（即子网络）。实验结果表明，从超级网络继承的权重的发现的结构达到了4.67 \％f1-able kitti上的误差，这是RAFT基线的8.4 \％降低，超过了先进的手工制作的型号GMA和AGFlow，同时降低模型的复杂性和延迟。源代码和训练有素的模型将在https://github.com/vdigpku/flownas中发布。