Modelpublic

skblv/yolov12m-cholect50-instruments

sourceHugging Faceagpl-3.0updated 24d agoView on Hugging Face

0likes40downloads

Model Card

YOLOv12-m — CholecT50 instrument recognition

Supervised YOLOv12-m detector trained for the 6 CholecT50 instruments, then scored as multi-label presence (a tool is present if any box for that class is predicted).

This is the “YOLOv12-m” row on the Instruments tab of the SDSC × Chicago Booth surgical video understanding leaderboard.

Prompt example

This closed-set example mirrors the leaderboard format, not a text-input API for this checkpoint.

text

[surgical frame]

Which instruments are visible in this laparoscopic cholecystectomy frame?
Select every matching label.

- grasper
- bipolar
- hook
- scissors
- clipper
- irrigator

Evaluation

Full CholecT50 validation split (95% bootstrap CI):

Metric	Value
Exact match	81.37% (80.87–81.92)
Micro-averaged F1	92.37% (92.11–92.60)

best.pt is the Ultralytics checkpoint; results_summary.json and per_tool_metrics.csv record the published eval.

License note

Derived from Ultralytics YOLO weights; distributed under AGPL-3.0.

References

—Skobelev, K., Fithian, E., Baranovski, Y., et al. A Comparative Study in Surgical AI: Potential and Limitations of Data, Compute, and Scaling. arXiv:2603.27341, 2026.
—Dataset: Nwoye, C. I., Yu, T., Gonzalez, C., et al. Rendezvous: Attention Mechanisms for the Recognition of Surgical Action Triplets in Endoscopic Videos. Medical Image Analysis, 78, 102433 (2022).

Limitations

Research baseline only. Not a medical device.