Dataset Versions

v4

2024-04-26 7:32pm

Generated on Apr 26, 2024

Dataset Split

Train Set 92%
604Images
Valid Set 5%
36Images
Test Set 3%
18Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640

Augmentations

Outputs per training example: 4
90° Rotate: Clockwise, Counter-Clockwise
Shear: ±10° Horizontal, ±10° Vertical
Grayscale: Apply to 15% of images
Saturation: Between -25% and +25%
Brightness: Between -21% and +21%
Exposure: Between -10% and +10%
Blur: Up to 4.9px
Noise: Up to 0.1% of pixels