dazinovic/neural-rgbd-surface-reconstruction
Neural radiance field approach for 3D surface reconstruction from RGB-D image sequences.

Velocity · 7d
+0.3
★ / day
Trend
→steady
star history
This repository implements a CVPR 2022 paper on neural RGB-D surface reconstruction. It combines implicit surface representations with neural radiance fields to perform 3D reconstruction from depth and color images. The code processes RGB-D data to optimize a scene representation, using deep learning techniques to handle noisy depth inputs and produce high-quality 3D models.