Optical Flow as Spatial-Temporal Attention Learners

Yawen Lu;Cheng Han;Qifan Wang;Heng Fan;Zhaodan Kong;Dongfang Liu;Yingjie Chen
{"title":"Optical Flow as Spatial-Temporal Attention Learners","authors":"Yawen Lu;Cheng Han;Qifan Wang;Heng Fan;Zhaodan Kong;Dongfang Liu;Yingjie Chen","doi":"10.1109/TPAMI.2024.3463648","DOIUrl":null,"url":null,"abstract":"Optical flow is an indispensable building block for various important computer vision tasks, including motion estimation, object tracking, and disparity measurement. To date, the dominant methods are CNN-based, leaving plenty of room for improvement. In this work, we propose TransFlow, a transformer architecture for optical flow estimation. Compared to dominant CNN-based methods, TransFlow demonstrates three advantages. First, it provides more accurate correlation and trustworthy matching in flow estimation by utilizing spatial self-attention and cross-attention mechanisms between adjacent frames to effectively capture global dependencies; Second, it recovers more compromised information (e.g., occlusion and motion blur) in flow estimation through long-range temporal association in dynamic scenes; Third, it introduces a concise self-learning paradigm, eliminating the need for complex and laborious multi-stage pre-training procedures. The versatility and superiority of TransFlow extend seamlessly to 3D scene motion, yielding competitive outcomes in 3D scene flow estimation. Our approach attains state-of-the-art results on benchmark datasets such as Sintel and KITTI-15, while also exhibiting exceptional performance on downstream tasks, including video object detection using the ImageNet VID dataset, video frame interpolation using the GoPro dataset, and video stabilization using the DeepStab dataset. We believe that the effectiveness of TransFlow positions it as a flexible baseline for both optical flow and scene flow estimation, offering promising avenues for future research and development.","PeriodicalId":94034,"journal":{"name":"IEEE transactions on pattern analysis and machine intelligence","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE transactions on pattern analysis and machine intelligence","FirstCategoryId":"1085","ListUrlMain":"https://ieeexplore.ieee.org/document/10705000/","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

Abstract

Optical flow is an indispensable building block for various important computer vision tasks, including motion estimation, object tracking, and disparity measurement. To date, the dominant methods are CNN-based, leaving plenty of room for improvement. In this work, we propose TransFlow, a transformer architecture for optical flow estimation. Compared to dominant CNN-based methods, TransFlow demonstrates three advantages. First, it provides more accurate correlation and trustworthy matching in flow estimation by utilizing spatial self-attention and cross-attention mechanisms between adjacent frames to effectively capture global dependencies; Second, it recovers more compromised information (e.g., occlusion and motion blur) in flow estimation through long-range temporal association in dynamic scenes; Third, it introduces a concise self-learning paradigm, eliminating the need for complex and laborious multi-stage pre-training procedures. The versatility and superiority of TransFlow extend seamlessly to 3D scene motion, yielding competitive outcomes in 3D scene flow estimation. Our approach attains state-of-the-art results on benchmark datasets such as Sintel and KITTI-15, while also exhibiting exceptional performance on downstream tasks, including video object detection using the ImageNet VID dataset, video frame interpolation using the GoPro dataset, and video stabilization using the DeepStab dataset. We believe that the effectiveness of TransFlow positions it as a flexible baseline for both optical flow and scene flow estimation, offering promising avenues for future research and development.
光流作为时空注意力学习器
光流是各种重要计算机视觉任务(包括运动估计、物体跟踪和差异测量)不可或缺的组成部分。迄今为止,主流的方法都是基于 CNN 的,因此还有很大的改进空间。在这项工作中,我们提出了用于光流估计的变换器架构 TransFlow。与基于 CNN 的主流方法相比,TransFlow 具有三个优势。首先,它利用相邻帧之间的空间自注意和交叉注意机制,有效捕捉全局依赖关系,从而在光流估计中提供更准确的相关性和可信匹配;其次,它通过动态场景中的长程时间关联,在光流估计中恢复更多受损信息(如遮挡和运动模糊);第三,它引入了简洁的自学习范式,无需复杂费力的多阶段预训练程序。TransFlow 的多功能性和优越性可无缝扩展到三维场景运动,从而在三维场景流量估算中获得极具竞争力的结果。我们的方法在 Sintel 和 KITTI-15 等基准数据集上取得了最先进的结果,同时在下游任务上也表现出卓越的性能,包括使用 ImageNet VID 数据集进行视频对象检测、使用 GoPro 数据集进行视频帧插值以及使用 DeepStab 数据集进行视频稳定。我们相信,TransFlow 的有效性使其成为光流和场景流估算的灵活基准,为未来的研究和发展提供了广阔的前景。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信