Unsupervised Point Cloud Pre-training via Occlusion Completion

By Hanchen Wang, Qi Liu, Xiangyu Yue, Joan Lasenby, Matt J. Kusner. IEEE International Conference on Computer Vision, 2020

336 citations recorded by the source corpus.

We describe a simple pre-training approach for point clouds. It works in three steps: 1. Mask all points occluded in a camera view; 2. Learn an encoder-decoder model to reconstruct the occluded points; 3. Use the encoder weights as initialisation for downstream point cloud tasks. We find that even when we pre-train on a single dataset (ModelNet40), this method improves accuracy across different datasets and encoders, on a wide range of downstream tasks. Specifically, we show that our method outperforms previous pre-training methods in object classification, and both part-based and semantic segmentation tasks. We study the pre-trained features and find that they lead to wide downstream minima, have high transformation invariance, and have activations that are highly correlated with part labels. Code and data are available at: https://github.com/hansen7/OcCo

Subfields: Aerospace Engineering, Computational Mechanics, Geology

View Paper: https://www.semanticscholar.org/paper/47d22522c9268ae597bf426f5bd14412575091a1

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

1. Hongsheng Li
2. Jie Zhou
3. Jiwen Lu
4. Yu Qiao
5. Rongyao Fang
6. Peng Gao
7. Bin Zhao
8. Renrui Zhang

In Computational Mechanics

1. Jie Zhou
2. Jiwen Lu
3. Yu Qiao
4. Rongyao Fang
5. Peng Gao
6. Lulu Tang
7. Bin Zhao
8. Renrui Zhang

In Geology

1. Jie Zhou
2. Hongsheng Li
3. Jiwen Lu
4. Yu Qiao
5. Peng Gao
6. Lulu Tang
7. Renrui Zhang
8. Xumin Yu

Citations in the corpus

• Masked Autoencoders for Point Cloud Self-supervised Learning (2022, methodology): “SSL for Point Cloud SSL has also been widely studied for point cloud representation learning [41,58,48,36,1,20,52,34,13,51].”
• Point-BERT: Pre-training 3D Point Cloud Transformers with Masked Point Modeling (2021, methodology): “More recently, OcCo [50] describes an encoder-decoder mechanism to reconstruct the occluded point clouds.”
• Point-M2AE: Multi-scale Masked Autoencoders for Hierarchical Point Cloud Pre-training (2022, methodology): “…[52] 83.3 Latent-GAN [46] 85.7 SO-Net [25] 87.3 FoldingNet [57] 88.4 MAP-VAE [19] 88.4 VIP-GAN [18] 90.2 DGCNN + Jiasaw [43] 90.6 DGCNN + OcCo [49] 90.7 DGCNN + CrossPoint [2] 91.2 Transformer + OcCo [60] 89.6 Point-BERT [60] 87.4 Point-M2AE 92.9 Improvement +1.7 1 , the extra spatial noises…”