Dataset Versions

v1

2023-07-02 5:13pm

Generated on Jul 2, 2023

Dataset Split

Train Set 82%
3096Images
Valid Set 9%
340Images
Test Set 9%
344Images

Preprocessing

Auto-Orient: Applied
Static Crop: 0-95% Horizontal Region, 20-66% Vertical Region
Resize: Stretch to 640x640
Tile: 2 rows x 2 columns

Augmentations

Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Saturation: Between -10% and +10%
Brightness: Between -30% and +30%