TY - GEN
T1 - FH-Net
T2 - 17th European Conference on Computer Vision, ECCV 2022
AU - Ding, Lihe
AU - Dong, Shaocong
AU - Xu, Tingfa
AU - Xu, Xinli
AU - Wang, Jie
AU - Li, Jianan
N1 - Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2022
Y1 - 2022
N2 - Estimating scene flow from real-world point clouds is a fundamental task for practical 3D vision. Previous methods often rely on deep models to first extract expensive per-point features at full resolution, and then get the flow either from complex matching mechanism or feature decoding, suffering high computational cost and latency. In this work, we propose a fast hierarchical network, FH-Net, which directly gets the key points flow through a lightweight Trans-flow layer utilizing the reliable local geometry prior, and optionally back-propagates the computed sparse flows through an inverse Trans-up layer to obtain hierarchical flows at different resolutions. To focus more on challenging dynamic objects, we also provide a new copy-and-paste data augmentation technique based on dynamic object pairs generation. Moreover, to alleviate the chronic shortage of real-world training data, we establish two new large-scale datasets to this field by collecting lidar-scanned point clouds from public autonomous driving datasets and annotating the collected data through novel pseudo-labeling. Extensive experiments on both public and proposed datasets show that our method outperforms prior state-of-the-arts while running at least 7× faster at 113 FPS. Code and data are released at https://github.com/pigtigger/FH-Net.
AB - Estimating scene flow from real-world point clouds is a fundamental task for practical 3D vision. Previous methods often rely on deep models to first extract expensive per-point features at full resolution, and then get the flow either from complex matching mechanism or feature decoding, suffering high computational cost and latency. In this work, we propose a fast hierarchical network, FH-Net, which directly gets the key points flow through a lightweight Trans-flow layer utilizing the reliable local geometry prior, and optionally back-propagates the computed sparse flows through an inverse Trans-up layer to obtain hierarchical flows at different resolutions. To focus more on challenging dynamic objects, we also provide a new copy-and-paste data augmentation technique based on dynamic object pairs generation. Moreover, to alleviate the chronic shortage of real-world training data, we establish two new large-scale datasets to this field by collecting lidar-scanned point clouds from public autonomous driving datasets and annotating the collected data through novel pseudo-labeling. Extensive experiments on both public and proposed datasets show that our method outperforms prior state-of-the-arts while running at least 7× faster at 113 FPS. Code and data are released at https://github.com/pigtigger/FH-Net.
KW - Copy-and-paste
KW - Real-world point cloud
KW - Scene flow
KW - Transformer
UR - http://www.scopus.com/inward/record.url?scp=85142688074&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-19842-7_13
DO - 10.1007/978-3-031-19842-7_13
M3 - Conference contribution
AN - SCOPUS:85142688074
SN - 9783031198410
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 213
EP - 229
BT - Computer Vision – ECCV 2022 - 17th European Conference, Proceedings
A2 - Avidan, Shai
A2 - Brostow, Gabriel
A2 - Cissé, Moustapha
A2 - Farinella, Giovanni Maria
A2 - Hassner, Tal
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 23 October 2022 through 27 October 2022
ER -