AMap open-sources ABot-Recon: streaming 3D reconstruction from video with a 12-frame local context
rsasaki0109 · x · 2026-10-02
ABot-Recon from amap-cvlab performs streaming 3D reconstruction from raw video, using a fixed 12-frame local context—composing current-frame geometry and adjacent relative poses into a global reconstruction—without persistent learned long-range memory. The repo (1.1k stars) includes a tech report, and fine-tuning code with 18 data sources, mixing/EMA/resume was recently released on the train branch.
More from Research
- Scientific superintelligence won't look like a smarter chatbot — bravo_abad · 2026-10-02
- Climate Change AI launches Projects Directory with 100+ AI-for-climate projects — anselm · 2026-10-02
- NJU's OneStreamer: 4B Streaming Video LLM Tops All Eight Streaming Benchmarks — NJU · 2026-10-02
- AgSpec Speeds Up Coding Agents Up to 4.76x with Retrieval-Based Speculative Decoding — Sumin Lee · 2026-10-02
- VTR-Bench: Best of 11 Video Models Still Hits 0.25 WER Rendering Scene Text — Yu Huang · 2026-10-02
- E-MoE Improves Few-Step Diffusion LMs Using Expert Routing as Shared Latent — Arseny Ivanov · 2026-10-02