# Silhouette Guided Point Cloud Reconstruction beyond Occlusion

## TL;DR

This paper introduces a silhouette-guided method for 3D point cloud reconstruction from a single RGB image, effectively handling occlusions and improving shape prediction accuracy.

## Contribution

It presents a novel silhouette completion and guidance approach that enhances 3D reconstruction robustness beyond occlusion, with additional improvements using multi-view reprojection and surface refinement.

## Key findings

- Improved reconstruction accuracy on synthetic and real datasets.
- Effective silhouette completion for occluded objects.
- Enhanced 3D shape prediction with multi-view loss and refinement.

## Abstract

One major challenge in 3D reconstruction is to infer the complete shape geometry from partial foreground occlusions. In this paper, we propose a method to reconstruct the complete 3D shape of an object from a single RGB image, with robustness to occlusion. Given the image and a silhouette of the visible region, our approach completes the silhouette of the occluded region and then generates a point cloud. We show improvements for reconstruction of non-occluded and partially occluded objects by providing the predicted complete silhouette as guidance. We also improve state-of-the-art for 3D shape prediction with a 2D reprojection loss from multiple synthetic views and a surface-based smoothing and refinement step. Experiments demonstrate the efficacy of our approach both quantitatively and qualitatively on synthetic and real scene datasets.