CUED Publications database

Unsupervised Point Cloud Pre-Training via View-Point Occlusion, Completion

Wang, H and Liu, Q and Yue, X and Lasenby, J and Kusner, MJ Unsupervised Point Cloud Pre-Training via View-Point Occlusion, Completion. (Unpublished)

Full text not available from this repository.

Abstract

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 construct a single pre-training dataset (from ModelNet40), this pre-training 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

Item Type: Article
Uncontrolled Keywords: cs.CV cs.CV cs.LG
Subjects: UNSPECIFIED
Divisions: Div F > Signal Processing and Communications
Depositing User: Cron Job
Date Deposited: 16 Oct 2020 20:11
Last Modified: 22 Apr 2021 05:02
DOI: