Dataset Versions

v16

2022-12-07 10:28pm

Generated on Dec 7, 2022

Dataset Split

Train Set 91%
6861Images
Valid Set 8%
617Images
Test Set 1%
71Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640

Augmentations

Outputs per training example: 3
Rotation: Between -20° and +20°
Hue: Between -43° and +43°
Saturation: Between -30% and +30%
Brightness: Between -46% and +46%
Exposure: Between -25% and +25%
Blur: Up to 1.75px
Noise: Up to 1% of pixels