Unpaired Point Cloud Completion on Real Scans using Adversarial Training

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.

arXiv commentICLR 2020

References in corpus (2)

• Learning Representations and Generative Models for 3D Point Clouds
• PCPNET: Learning Local Shape Properties from Raw Point Clouds

Cited by in corpus (13)

• Density-aware Chamfer Distance as a Comprehensive Metric for Point Cloud Completion
• Self-Supervised Learning for Domain Adaptation on Point-Clouds
• SPG: Unsupervised Domain Adaptation for 3D Object Detection via Semantic Point Generation
• Unsupervised 3D Shape Completion through GAN Inversion
• Cycle4Completion: Unpaired Point Cloud Completion using Cycle Transformation with Missing Region Coding
• Deep Mesh Prior: Unsupervised Mesh Restoration using Graph Convolutional Networks
• Multimodal Shape Completion via Conditional Generative Adversarial Networks
• Point Cloud Upsampling via Disentangled Refinement
• SnowflakeNet: Point Cloud Completion by Snowflake Point Deconvolution with Skip-Transformer
• Generation For Adaption: A GAN-Based Approach for 3D Domain Adaption with Point Cloud Data
• Augmented Environment Representations with Complete Object Models
• A Convolutional Architecture for 3D Model Embedding
• Cascaded Refinement Network for Point Cloud Completion with Self-supervision