Atlas
An end-to-end deep learning system for reconstructing complete 3D scenes (geometry and semantics) from posed 2D images.
What is Atlas?
Atlas is an end-to-end deep learning system for 3D scene reconstruction from posed 2D images. It takes a set of images with known camera poses and outputs a complete 3D scene representation including both geometry (as a mesh or TSDF) and semantic segmentation per vertex. It solves the problem of creating detailed, semantically-aware 3D models from image collections without relying on traditional multi-view stereo pipelines.
Computer vision researchers and engineers working on 3D reconstruction, SLAM, AR/VR, and robotics who need to generate semantically labeled 3D environments from images.
Atlas provides a fully learned pipeline that can produce more complete and semantically meaningful reconstructions compared to traditional methods like COLMAP, especially in challenging areas with textureless surfaces or occlusions, by leveraging deep learning priors trained on large datasets.
Overview
Atlas: End-to-End 3D Scene Reconstruction from Posed Images
Use Cases
Best For
- Reconstructing indoor scenes with semantic labels (e.g., walls, floors, furniture)
Related Projects
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
