Unpaired Point Cloud Completion on Real Scans using Adversarial Training

TL;DR

This paper introduces an adversarial training method for unpaired point cloud completion, enabling realistic 3D scan completion directly on real-world data without needing paired training samples.

Contribution

It presents the first unpaired, real-scan compatible point cloud completion approach using adversarial training, bypassing the need for paired datasets.

Findings

Effective on real-world datasets like ScanNet, Matterport, and KITTI.

Achieves realistic completions with varying levels of incompleteness.

Outperforms existing methods on the 3D-EPN benchmark.

Abstract

As 3D scanning solutions become increasingly popular, several deep learning setups have been developed geared towards that task of scan completion, i.e., plausibly filling in regions there were missed in the raw scans. These methods, however, largely rely on supervision in the form of paired training data, i.e., partial scans with corresponding desired completed scans. While these methods have been successfully demonstrated on synthetic data, the approaches cannot be directly used on real scans in absence of suitable paired training data. We develop a first approach that works directly on input point clouds, does not require paired training data, and hence can directly be applied to real scans for scan completion. We evaluate the approach qualitatively on several real-world datasets (ScanNet, Matterport, KITTI), quantitatively on 3D-EPN shape completion benchmark dataset, and demonstrate realistic completions under varying levels of incompleteness.