Dataset Versions

v10

2023-10-24 7:32pm

Generated on Oct 24, 2023

Dataset Split

Train Set 82%
3648Images
Valid Set 12%
532Images
Test Set 6%
256Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640
Modify Classes: 8 remapped, 0 dropped

Augmentations

Outputs per training example: 2
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Shear: ±27° Horizontal, ±30° Vertical
Noise: Up to 10% of pixels
Cutout: 6 boxes with 5% size each