Views
No views yet
| id | class | meaning |
|---|---|---|
| 0 | doorway | passable opening — open/ajar door (incl. open glass door), framed hole you can walk through |
| 1 | door | closed door slab (incl. closed glass door) |
| 2 | people | any person |
| 3 | window | windows / glass surfaces (glass walls, panels) — not mirrors |
mirror is internal: it is detected but never output — used at inference to
suppress door/doorway/window predictions that fall on a mirror (reflections are
phantom obstacles). Run predictions through mirror_suppress.finalize().door: ~79–82% (conference-room case).doorway_seg_yolo26n.pt — Ultralytics weights (train/val/predict).doorway_seg_yolo26n.onnx — NMS-free ONNX (opset 17, 640×640) for TensorRT/ONNXRuntime; build the Jetson engine on-device with trtexec.mirror_suppress.py — inference post-process: mirror suppression + portal de-dup (finalize).predict.py — demo: run on image/video/webcam with the post-process applied.1from ultralytics import YOLO
2from mirror_suppress import finalize, dets_from_result # from this repo
3
4model = YOLO("doorway_seg_yolo26n.pt")
5r = model.predict("frame.jpg", conf=0.10, imgsz=640)[0] # predict low; finalize re-thresholds
6dets = finalize(dets_from_result(r), out_conf=0.35) # 4-class output, reflections dropped
7# dets: [{"cls":0..3, "box":[x1,y1,x2,y2], "conf":...}, ...]mirror class. Quantization: FP16 lossless; INT8 PTQ ≈ −3 mAP pts.doorway is the rarest/hardest class (small AP); open passages viewed straight-on can read as door.