Dataset Versions

2023-06-15 11:46pm

Generated on Jun 15, 2023

Train Set 87%

396Images

Valid Set 8%

37Images

Test Set 4%

20Images

Auto-Orient: Applied

Resize: Stretch to 640x640

Auto-Adjust Contrast: Using Adaptive Equalization

Outputs per training example: 3

Flip: Horizontal, Vertical

Rotation: Between -20° and +20°

Saturation: Between -48% and +48%

Blur: Up to 2px

Cutout: 3 boxes with 10% size each

Bounding Box: Shear: ±15° Horizontal, ±15° Vertical

Bounding Box: Noise: Up to 5% of pixels