YOLOv8 Bolt Quality Detection
Custom-trained YOLOv8 model for detecting and classifying bolts in industrial manufacturing contexts.
Classes
Boulon_Bon— conforming bolt (green bounding box)Boulon_Mauvais— non-conforming / defective bolt (red bounding box)
Training
- Dataset: 1,322 annotated images (Roboflow + Kaggle + custom)
- Split: 70% train / 20% validation / 10% test
- Epochs: 200
- Hardware: Google Colab T4 GPU
- Framework: Ultralytics YOLOv8
Usage
Download best.pt and run with the full application: https://github.com/yassir28ab/bolt-quality-detection