TY - GEN
T1 - Canonical Voting
T2 - 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022
AU - You, Yang
AU - Ye, Zelin
AU - Lou, Yujing
AU - Li, Chengkun
AU - Li, Yong Lu
AU - Ma, Lizhuang
AU - Wang, Weiming
AU - Lu, Cewu
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - 3D object detection has attracted much attention thanks to the advances in sensors and deep learning methods for point clouds. Current state-of-the-art methods like VoteNet regress direct offset towards object centers and box orientations with an additional Multi-Layer-Perceptron network. Both their offset and orientation predictions are not accurate due to the fundamental difficulty in rotation classification. In the work, we disentangle the direct offset into Local Canonical Coordinates (LCC), box scales and box orientations. Only LCC and box scales are regressed, while box orientations are generated by a canonical voting scheme. Finally, an LCC-aware back-projection checking algorithm iteratively cuts out bounding boxes from the generated vote maps, with the elimination of false positives. Our model achieves state-of-the-art performance on three standard real-world benchmarks: ScanNet, SceneNN and SUN RGB-D. Our code is available on https://github.com/qq456cvb/CanonicalVoting.
AB - 3D object detection has attracted much attention thanks to the advances in sensors and deep learning methods for point clouds. Current state-of-the-art methods like VoteNet regress direct offset towards object centers and box orientations with an additional Multi-Layer-Perceptron network. Both their offset and orientation predictions are not accurate due to the fundamental difficulty in rotation classification. In the work, we disentangle the direct offset into Local Canonical Coordinates (LCC), box scales and box orientations. Only LCC and box scales are regressed, while box orientations are generated by a canonical voting scheme. Finally, an LCC-aware back-projection checking algorithm iteratively cuts out bounding boxes from the generated vote maps, with the elimination of false positives. Our model achieves state-of-the-art performance on three standard real-world benchmarks: ScanNet, SceneNN and SUN RGB-D. Our code is available on https://github.com/qq456cvb/CanonicalVoting.
KW - RGBD sensors and analytics
KW - Recognition: detection
KW - Scene analysis and understanding
KW - Vision applications and systems
KW - categorization
KW - retrieval
UR - https://www.scopus.com/pages/publications/85126961521
U2 - 10.1109/CVPR52688.2022.00126
DO - 10.1109/CVPR52688.2022.00126
M3 - 会议稿件
AN - SCOPUS:85126961521
T3 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
SP - 1183
EP - 1192
BT - Proceedings - 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022
PB - IEEE Computer Society
Y2 - 19 June 2022 through 24 June 2022
ER -