speed3r
[CVPR 2026 Findings] Speed3R: Sparse Feed-forward 3D Reconstruction Models
Take a look →OPEN SOURCE, IN FOCUS
jndeng/saf3r
[ECCV 2026] Dynamic Sparse Attention for Feed-Forward 3D Reconstruction Transformers
Saved projects stay in this browser. No account needed.
TAKE A CLOSER LOOK
The repository is the place to check installation steps, supported platforms, the license, and current maintenance before choosing a tool.
Read the documentation on GitHub ↗Find projects in the same areas, or search the archive for a different approach.