Asymmetric convolution kernel for deep optical flow estimation

Yueran Zu; Wenzhong Tang; Xiuguo Bao; Yanyang Wang; Shuai Wang

doi:10.1109/AEECA49918.2020.9213629

Asymmetric convolution kernel for deep optical flow estimation

Yueran Zu
, Wenzhong Tang
, Xiuguo Bao^*
, Yanyang Wang
, Shuai Wang

^*此作品的通讯作者

Beihang University
National Computer Network Emergency Response Technical

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

1 引用（Scopus）

摘要

Deep optical flow estimation is a hot topic in computer vision in recent years. In this paper, an asymmetric convolution kernel is introduced to improve the accuracy of optical flow. Then, in order to alleviate the rasterization problem caused by dilated convolution in optical flow results, a method of hybrid dilated convolution is introduced. At the same time, the proposed method can keep a small number of parameters without using additional occlusion and bidirectional information. Finally, experiments on the open standard datasets MPI-Sintel and KITTI-15 are carried out, and the results demonstrate the effectiveness of the proposed approach.

源语言	英语
主期刊名	Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020
出版商	Institute of Electrical and Electronics Engineers Inc.
页	266-270
页数	5
ISBN（电子版）	9781728165202
DOI	https://doi.org/10.1109/AEECA49918.2020.9213629
出版状态	已出版 - 8月 2020
活动	2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020 - Dalian, 中国期限: 25 8月 2020 → 27 8月 2020

出版系列

姓名	Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020

会议

会议	2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020
国家/地区	中国
市	Dalian
时期	25/08/20 → 27/08/20

访问文件

10.1109/AEECA49918.2020.9213629

其它文件与链接

链接到 Scopus 的出版物

引用此

Zu, Y., Tang, W., Bao, X., Wang, Y., & Wang, S. (2020). Asymmetric convolution kernel for deep optical flow estimation. 在 Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020 (页码 266-270). 文章 9213629 (Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/AEECA49918.2020.9213629

Zu, Yueran ; Tang, Wenzhong ; Bao, Xiuguo 等. / Asymmetric convolution kernel for deep optical flow estimation. Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020. Institute of Electrical and Electronics Engineers Inc., 2020. 页码 266-270 (Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020).

@inproceedings{6263db4c77264334802f23d3f839a230,

title = "Asymmetric convolution kernel for deep optical flow estimation",

abstract = "Deep optical flow estimation is a hot topic in computer vision in recent years. In this paper, an asymmetric convolution kernel is introduced to improve the accuracy of optical flow. Then, in order to alleviate the rasterization problem caused by dilated convolution in optical flow results, a method of hybrid dilated convolution is introduced. At the same time, the proposed method can keep a small number of parameters without using additional occlusion and bidirectional information. Finally, experiments on the open standard datasets MPI-Sintel and KITTI-15 are carried out, and the results demonstrate the effectiveness of the proposed approach.",

keywords = "asymmetric convolution kernel, deep learning, dilated convolution, optical flow estimation",

author = "Yueran Zu and Wenzhong Tang and Xiuguo Bao and Yanyang Wang and Shuai Wang",

note = "Publisher Copyright: {\textcopyright} 2020 IEEE.; 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020 ; Conference date: 25-08-2020 Through 27-08-2020",

year = "2020",

month = aug,

doi = "10.1109/AEECA49918.2020.9213629",

language = "英语",

series = "Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "266--270",

booktitle = "Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020",

address = "美国",

}

Zu, Y, Tang, W, Bao, X, Wang, Y & Wang, S 2020, Asymmetric convolution kernel for deep optical flow estimation. 在 Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020., 9213629, Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020, Institute of Electrical and Electronics Engineers Inc., 页码 266-270, 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020, Dalian, 中国, 25/08/20. https://doi.org/10.1109/AEECA49918.2020.9213629

Asymmetric convolution kernel for deep optical flow estimation. / Zu, Yueran; Tang, Wenzhong; Bao, Xiuguo 等.
Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020. Institute of Electrical and Electronics Engineers Inc., 2020. 页码 266-270 9213629 (Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020).

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

TY - GEN

T1 - Asymmetric convolution kernel for deep optical flow estimation

AU - Zu, Yueran

AU - Tang, Wenzhong

AU - Bao, Xiuguo

AU - Wang, Yanyang

AU - Wang, Shuai

PY - 2020/8

Y1 - 2020/8

N2 - Deep optical flow estimation is a hot topic in computer vision in recent years. In this paper, an asymmetric convolution kernel is introduced to improve the accuracy of optical flow. Then, in order to alleviate the rasterization problem caused by dilated convolution in optical flow results, a method of hybrid dilated convolution is introduced. At the same time, the proposed method can keep a small number of parameters without using additional occlusion and bidirectional information. Finally, experiments on the open standard datasets MPI-Sintel and KITTI-15 are carried out, and the results demonstrate the effectiveness of the proposed approach.

AB - Deep optical flow estimation is a hot topic in computer vision in recent years. In this paper, an asymmetric convolution kernel is introduced to improve the accuracy of optical flow. Then, in order to alleviate the rasterization problem caused by dilated convolution in optical flow results, a method of hybrid dilated convolution is introduced. At the same time, the proposed method can keep a small number of parameters without using additional occlusion and bidirectional information. Finally, experiments on the open standard datasets MPI-Sintel and KITTI-15 are carried out, and the results demonstrate the effectiveness of the proposed approach.

KW - asymmetric convolution kernel

KW - deep learning

KW - dilated convolution

KW - optical flow estimation

UR - https://www.scopus.com/pages/publications/85094679990

U2 - 10.1109/AEECA49918.2020.9213629

DO - 10.1109/AEECA49918.2020.9213629

M3 - 会议稿件

AN - SCOPUS:85094679990

T3 - Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020

SP - 266

EP - 270

BT - Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020

Y2 - 25 August 2020 through 27 August 2020

ER -

Zu Y, Tang W, Bao X, Wang Y , Wang S. Asymmetric convolution kernel for deep optical flow estimation. 在 Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020. Institute of Electrical and Electronics Engineers Inc. 2020. 页码 266-270. 9213629. (Proceedings of 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications, AEECA 2020). doi: 10.1109/AEECA49918.2020.9213629

Asymmetric convolution kernel for deep optical flow estimation

摘要

出版系列

会议

访问文件

其它文件与链接

指纹

引用此