/occupancy_networks

This repository contains the code for the paper "Occupancy Networks - Learning 3D Reconstruction in Function Space"

Primary LanguagePythonMIT LicenseMIT

Occupancy Networks

Several attempts to improve the performance of ONet on the task of single-image reconstruction.

Currently, by introducing depth map & local feature extraction into the single-image reconstruction pipeline, the mesh_iou (evaluated on the original 100000 points) can achieve 65.1%, which is better than 59.7% from the original ONet. The reconstructions are also visually better.

I'm trying hard to write a paper for these contributions :).