Dataset Versions

v1

2024-07-28 4:48pm

Generated on Jul 28, 2024

Dataset Split

Train Set 90%
5296Images
Valid Set 6%
376Images
Test Set 3%
194Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640

Augmentations

Outputs per training example: 4
Rotation: Between -20° and +20°
Grayscale: Apply to 25% of images
Saturation: Between -25% and +25%
Brightness: Between -25% and +25%
Exposure: Between -12% and +12%
Blur: Up to 3.3px
Noise: Up to 4.65% of pixels