TY - GEN
T1 - HybridCR
T2 - 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022
AU - Li, Mengtian
AU - Xie, Yuan
AU - Shen, Yunhang
AU - Ke, Bo
AU - Qiao, Ruizhi
AU - Ren, Bo
AU - Lin, Shaohui
AU - Ma, Lizhuang
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - To address the huge labeling cost in large-scale point cloud semantic segmentation, we propose a novel hybrid contrastive regularization (HybridCR) framework in weakly-supervised setting, which obtains competitive performance compared to its fully-supervised counterpart. Specifically, HybridCR is the first framework to leverage both point consistency and employ contrastive regularization with pseudo labeling in an end-to-end manner. Fundamentally, HybridCR explicitly and effectively considers the semantic similarity between local neighboring points and global characteristics of 3D classes. We further design a dynamic point cloud augmentor to generate diversity and robust sample views, whose transformation parameter is jointly optimized with model training. Through extensive experiments, HybridCR achieves significant performance improvement against the SOTA methods on both indoor and outdoor datasets, e.g., S3DIS, ScanNet-V2, Semantic3D, and SemanticKITTI.
AB - To address the huge labeling cost in large-scale point cloud semantic segmentation, we propose a novel hybrid contrastive regularization (HybridCR) framework in weakly-supervised setting, which obtains competitive performance compared to its fully-supervised counterpart. Specifically, HybridCR is the first framework to leverage both point consistency and employ contrastive regularization with pseudo labeling in an end-to-end manner. Fundamentally, HybridCR explicitly and effectively considers the semantic similarity between local neighboring points and global characteristics of 3D classes. We further design a dynamic point cloud augmentor to generate diversity and robust sample views, whose transformation parameter is jointly optimized with model training. Through extensive experiments, HybridCR achieves significant performance improvement against the SOTA methods on both indoor and outdoor datasets, e.g., S3DIS, ScanNet-V2, Semantic3D, and SemanticKITTI.
KW - Computer vision for social good
KW - Segmentation
KW - Self-& semi-& meta- Transfer/low-shot/long-tail learning
KW - grouping and shape analysis
UR - https://www.scopus.com/pages/publications/85137703525
U2 - 10.1109/CVPR52688.2022.01451
DO - 10.1109/CVPR52688.2022.01451
M3 - 会议稿件
AN - SCOPUS:85137703525
T3 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
SP - 14910
EP - 14919
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 -