Dataset Versions

v5

2024-02-13 12:51am

Generated on Feb 13, 2024

Dataset Split

Train Set 92%
820Images
Valid Set 5%
48Images
Test Set 3%
25Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640

Augmentations

Outputs per training example: 5
Rotation: Between -9° and +9°
Hue: Between -7° and +7°
Saturation: Between -8% and +8%
Brightness: Between -10% and +10%
Exposure: Between -5% and +5%
Blur: Up to 0.2px
Noise: Up to 0.47% of pixels