hinged-state-classifier-yolo26s-cls

YOLO26s classification model for surgical instrument state recognition.

Task

Binary classification of hinged surgical instrument state:

• closed
• open

Model details

• Architecture: YOLO26s-cls
• Framework: Ultralytics YOLO
• Checkpoint: best.pt
• Input image size: 224 × 224
• Epochs: 40
• Batch size: 32
• Device used for training: MPS
• Pretrained: yes
• Optimizer: auto
• Initial learning rate (lr0): 0.01
• Weight decay: 0.0005

Dataset

Trained on:

• joonhaim/hinged-state-classifier-crops

Notes

This is a custom Ultralytics YOLO classification checkpoint for classifying the state of hinged surgical instruments from cropped images.

The class order must match the order used during training.