โ† sources

๐๐ฎ๐ข๐ฅ๐๐ข๐ง๐  ๐š ๐ซ๐ž๐š๐ฅ-๐ญ๐ข๐ฆ๐ž ๐ง๐š๐ฏ๐ข๐ ๐š๐ญ๐ข๐จ๐ง ๐ฌ๐ฒ๐ฌ๐ญ๐ž๐ฆ ๐Ÿ๐จ๐ซ ๐ญ๐ก๐ž ๐ฏ๐ข๐ฌ๐ฎ๐š๐ฅ๐ฅ๐ฒ ๐ข๐ฆ๐ฉ๐š๐ข๐ซ๐ž๐ โ€” ๐š ๐Ÿ๐ž๐ฐ ๐ฐ๐ž๐ž๐ค๐ฌ ๐ข๐ง

source: reddit (reddit:computervision)published 2026-09-26 06:54ingested 2026-09-27 02:36

retained text

Started this a couple of weeks ago. No roadmap, just a problem worth solving.  
  
The idea: a system that processes what a camera sees and turns it into spoken directions a visually impaired person can actually act on. Not just "obstacle detected" โ€” but "Stairs on left, 1 metre. Stop."  
  
Navigation logic:  
Perspective trapezoid corridor model with bilateral (left/right) clearance analysis  
Hysteresis state machine: CLEAR โ†’ CAUTION โ†’ STOP, with instant escalation on imminent hazards  
Temporal smoothing across frames to kill jitter  
  
What's next: First prototype as a mobile app, then moving toward dedicated hardware.  
Still early. Still rough around the edges. But it works on real footage โ€” stairs, railings, curbs, moving through corridors.  
If you're working on assistive tech, accessibility, or computer vision โ€” I'd genuinely love to connect.

document fetched 2026-09-27 00:23 ยท canonical https://reddit.com/r/computervision/comments/1wqjqy9/%F0%9D%90%81%F0%9D%90%AE%F0%9D%90%A2%F0%9D%90%A5%F0%9D%90%9D%F0%9D%90%A2%F0%9D%90%A7%F0%9D%90%A0_%F0%9D%90%9A_%F0%9D%90%AB%F0%9D%90%9E%F0%9D%90%9A%F0%9D%90%A5%F0%9D%90%AD%F0%9D%90%A2%F0%9D%90%A6%F0%9D%90%9E_%F0%9D%90%A7%F0%9D%90%9A%F0%9D%90%AF%F0%9D%90%A2%F0%9D%90%A0%F0%9D%90%9A%F0%9D%90%AD%F0%9D%90%A2%F0%9D%90%A8%F0%9D%90%A7_%F0%9D%90%AC%F0%9D%90%B2%F0%9D%90%AC%F0%9D%90%AD%F0%9D%90%9E%F0%9D%90%A6_%F0%9D%90%9F%F0%9D%90%A8%F0%9D%90%AB_%F0%9D%90%AD%F0%9D%90%A1%F0%9D%90%9E