计算机与现代化Issue(2)：64-68,74,6.DOI:10.3969/j.issn.1006-2475.2024.02.010
基于反向残差注意力的光流估计
Optical Flow Estimation Based on Inverse Residual Attention
摘要
Abstract
Optical flow estimation is a basic task of video understanding and analysis.Many existing methods directly take occlu-sion as the outer point and eliminate it,so as to improve the ability of the model to calculate the optical flow,but it is also easy to cause the image gray discontinuity,leading to the failure of optical flow estimation.In addition,the problem of large displace-ment caused by high speed motion of objects has always been a difficulty in optical flow estimation.In order to solve the above problems,this paper proposes a generative adversarial learning framework based on reverse residual attention(FlowTranGAN,FTGAN)for optical flow estimation.The proposed framework enhances the spatial information of features by designing a reverse residual attention module to improve the matching degree between pixels.Besides,we use a discriminator based on U-Net to con-strain the generator to reduce the error and discontinuity of optical flow estimation,and improve the generalization ability of the model.Experiment results on the KITTI-2015 dataset and MPI-Sintel dataset demonstrate the effectiveness and superiority of the proposed FTGAN.关键词
光流估计/反向残差注意力/生成对抗学习/有监督学习Key words
optical flow estimation/reverse residual attention/generative adversarial learning/supervised learning分类
信息技术与安全科学引用本文复制引用
梁建业,陈俊洪,方桂标,吴兴财,刘文印..基于反向残差注意力的光流估计[J].计算机与现代化,2024,(2):64-68,74,6.基金项目
国家自然科学基金资助项目(91748107,62076073,61902077) （91748107,62076073,61902077）
广东省引进创新科研团队计划项目(2014ZT05G157) （2014ZT05G157）
广东省基础与应用基础研究基金资助项目(2020A1515010616) （2020A1515010616）
广东省科技创新战略专项资金资助项目(pdjh2020a0173) （pdjh2020a0173）
