SAM + DINOv3 turns a surgical tray into a local visual search engine
MaziyarPanahi · x · 2026-10-03
The author combined SAM 3.1 + DINOv3, fully local, to turn a single frame of a surgical instrument tray into a visual search engine: SAM isolates each instrument into masks, DINOv3 compares the content inside each mask by visual similarity. Using one instrument as the query, all others are ranked by similarity — no need to even know the instrument names. Next step: keeping the selection tracked as instruments move in video.
Related event: Dev Combines DINOv3 and SAM 3.1 into Fully Local Visual Search Engine(2 posts)→
More from coding & agent
- Dev ports NTSCRT retro effects to Godot using Opus 5.5, ships open-source addon — AIandDesign · 2026-10-03
- DavidKPiano open-sources Jevspresso: XState constrains LLM agent actions in a coffee-bar demo — DavidKPiano · 2026-10-03
- Top 10 trending Claude skills of the week: agent skills layer heats up — CurieuxExplorer · 2026-10-03
- AI Village dataset with millions of agent behavior samples trends on Hugging Face — aidigestorg · 2026-10-03
- Phone agent turn-taking: VAD plus transcript completeness beats silence timers alone — Kindly-Duty272 · 2026-10-03
- NYC Agentic AI meetup to recap September AI moves with live PowerShell decision demo — dfinke · 2026-10-03