NAVER open-sources BLASt3R: pose-free 3D reconstruction from any image set hits SOTA
kwangmoo_yi · x · 2026-09-09
NAVER researchers released BLASt3R, a 3D reconstruction system that performs bundle adjustment over arbitrary image sets using multi-view matching and monocular priors.
- Takes unposed, uncalibrated, unordered images or video and jointly recovers camera poses, intrinsics, and dense depthmaps per frame
- Offline mode handles unordered collections for maximum accuracy; online mode works like incremental SLAM on sequences
- Uses coarse/fine matching with depth scaling; the author reports state-of-the-art results
Code is open-sourced on GitHub (naver/blast3r), requiring a GPU with CUDA 12 and Python 3.11-3.13.
Related event: NAVER Open-Sources BLASt3R for Calibration-Free 3D Reconstruction(2 posts)→
More from Research
- Token Collapse: Study Finds Graph Soft Prompting Tokens in LLMs Fail to Encode Structure — Sapienza · 2026-10-27
- Mathathon 2.0 launches with Terence Tao's SAIR, $1.5M in compute for teams — _sathvikr · 2026-10-27
- The 'sweet lesson' is real: two kinds of ML problems need opposite strategies — mike64_t · 2026-10-27
- PixGR: Pixel-space generative decoding for fast feed-forward 3D reconstruction — kwangmoo_yi · 2026-10-27
- UniAction: unified action conditioning lets one world model handle mixed action types — ZeYanjie · 2026-10-27
- Infini-AI Lab open-sources Hermit, a framework for agent harness evolution and continual learning — BeidiChen · 2026-10-27