Reconstruction-Aware Prior Distillation for Semi-Supervised Point Cloud Completion
Abstract
Real-world sensors often produce incomplete, irregular, and noisy point clouds, making point cloud completion increasingly important. However, most existing completion methods rely on large paired datasets for training, which is labor-intensive. This paper proposes RaPD, a novel semi-supervised point cloud completion method that reduces the need for paired datasets. RaPD utilizes a two-stage training scheme, where a deep semantic prior is learned in stage 1 from unpaired complete and incomplete point clouds, and a semi-supervised prior distillation process is introduced in stage 2 to train a completion network using only a small number of paired samples. Additionally, a self-supervised completion module is introduced to improve performance using unpaired incomplete point clouds. Experiments on multiple datasets show that RaPD outperforms previous methods in both homologous and heterologous scenarios.
Cite
Text
Fan et al. "Reconstruction-Aware Prior Distillation for Semi-Supervised Point Cloud Completion." International Joint Conference on Artificial Intelligence, 2023. doi:10.24963/IJCAI.2023/81Markdown
[Fan et al. "Reconstruction-Aware Prior Distillation for Semi-Supervised Point Cloud Completion." International Joint Conference on Artificial Intelligence, 2023.](https://mlanthology.org/ijcai/2023/fan2023ijcai-reconstruction/) doi:10.24963/IJCAI.2023/81BibTeX
@inproceedings{fan2023ijcai-reconstruction,
title = {{Reconstruction-Aware Prior Distillation for Semi-Supervised Point Cloud Completion}},
author = {Fan, Zhaoxin and He, Yulin and Wang, Zhicheng and Wu, Kejian and Liu, Hongyan and He, Jun},
booktitle = {International Joint Conference on Artificial Intelligence},
year = {2023},
pages = {726-735},
doi = {10.24963/IJCAI.2023/81},
url = {https://mlanthology.org/ijcai/2023/fan2023ijcai-reconstruction/}
}