Point Cloud Pre-training with
Natural 3D Structures


Ryosuke Yamada1Hirokatsu Kataoka1Naoya Chiba2   Yasuyuki Domae1   Tetsuya Ogata1,2
1: AIST   2: Waseda Universsity




Abstract

The construction of 3D point cloud datasets requires a great deal of human effort. Therefore, constructing a large-scale 3D point clouds dataset is difficult. In order to remedy this issue, we propose a newly developed point cloud fractal database (PC-FractalDB), which is a novel family of formula-driven supervised learning inspired by fractal geometry encountered in natural 3D structures. Our research is based on the hypothesis that we could learn representations from more real-world 3D patterns than conventional 3D datasets by learning fractal geometry. We show how the PC-FractalDB facilitates solving several recent dataset-related problems in 3D scene understanding, such as 3D model collection and labor-intensive annotation. The experimental section shows how we achieved the performance rate of up to 61.9% and 59.4% for the ScanNetV2 and SUN RGB-D datasets, respectively, over the current highest scores obtained with the PointContrast, contrastive scene contexts (CSC), and RandomRooms. Moreover, the PC-FractalDB pre-trained model is especially effective in training with limited data. For example, in 10% of training data on ScanNetV2, the PC-FractalDB pre-trained VoteNet performs at 38.3%, which is +14.8% higher accuracy than CSC. Of particular note, we found that the proposed method achieves the highest results for 3D object detection pre-training in limited point cloud data.


Framework

Overview of the formula-driven supervised learning framework for 3D object detection with 3D point clouds. We generate a 3D fractal model using the 3D iterated function system. The proposed PC-FractalDB is automatically constructed by difiniting a fractal category using variance threshold and instance augmentation with FractalNoiseMix. A 3D fractal scene is generated by randomly selecting 3D fractal models and translating these from the origin on the z-plane.





Experimental Results

3D object detection comparisons on representative datasets. We employed architecture with the basic VoteNet model and used them to compare network pre-training methods, including training from scratch, PointContrast, CSC, RandomRooms, and the PC-FractalDB.





Additional Results

The PC-FractalDB pre-train can acquire effective features compared to previous self-supervised learning methods for limited data on fine-tuning datasets.



Citation

@InProceedings{Yamada_2022_CVPR, author = {Yamada, Ryosuke and Kataoka, Hirokatsu and Chiba, Naoya and Domae, Yukiyasu and Ogata, Tetsuya}, title = {Point Cloud Pre-Training With Natural 3D Structures}, booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, month = {June}, year = {2022}, pages = {21283-21293}

Acknowledgement