Dataset Versions

v3

2022-11-15 2:32pm

Generated on Nov 15, 2022

Dataset Split

Train Set 84%
102Images
Valid Set 12%
14Images
Test Set 4%
5Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 1000x1000

Augmentations

Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise
Grayscale: Apply to 25% of images
Hue: Between -34° and +34°
Saturation: Between -14% and +14%
Brightness: Between -25% and +25%
Blur: Up to 0.5px