360SD-Net: 360° Stereo Depth Estimation with Learnable Cost Volume

Ning Hsu Wang, Bolivar Solarte, Yi Hsuan Tsai, Wei Chen Chiu, Min Sun

研究成果: Conference contribution同行評審

4 引文 斯高帕斯(Scopus)

摘要

Recently, end-to-end trainable deep neural networks have significantly improved stereo depth estimation for perspective images. However, 360° images captured under equirectangular projection cannot benefit from directly adopting existing methods due to distortion introduced (i.e., lines in 3D are not projected onto lines in 2D). To tackle this issue, we present a novel architecture specifically designed for spherical disparity using the setting of top-bottom 360° camera pairs. Moreover, we propose to mitigate the distortion issue by (1) an additional input branch capturing the position and relation of each pixel in the spherical coordinate, and (2) a cost volume built upon a learnable shifting filter. Due to the lack of 360° stereo data, we collect two 360° stereo datasets from Matterport3D and Stanford3D for training and evaluation. Extensive experiments and ablation study are provided to validate our method against existing algorithms. Finally, we show promising results on real-world environments capturing images with two consumer-level cameras. Our project page is at https://albert100121.github.io/360SD-Net-Project-Page.

原文English
主出版物標題2020 IEEE International Conference on Robotics and Automation, ICRA 2020
發行者Institute of Electrical and Electronics Engineers Inc.
頁面582-588
頁數7
ISBN(電子)9781728173955
DOIs
出版狀態Published - 五月 2020
事件2020 IEEE International Conference on Robotics and Automation, ICRA 2020 - Paris, France
持續時間: 31 五月 202031 八月 2020

出版系列

名字Proceedings - IEEE International Conference on Robotics and Automation
ISSN(列印)1050-4729

Conference

Conference2020 IEEE International Conference on Robotics and Automation, ICRA 2020
國家/地區France
城市Paris
期間31/05/2031/08/20

指紋

深入研究「360SD-Net: 360° Stereo Depth Estimation with Learnable Cost Volume」主題。共同形成了獨特的指紋。

引用此