Dataset Versions

v3

2023-02-20 4:35pm

Generated on Feb 20, 2023

Dataset Split

Train Set 92%
735Images
Valid Set 8%
63Images
Test Set 0%
3Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640

Augmentations

Outputs per training example: 3
Rotation: Between -15° and +15°
Shear: ±8° Horizontal, ±12° Vertical
Grayscale: Apply to 25% of images
Hue: Between -70° and +70°
Exposure: Between -25% and +25%
Noise: Up to 5% of pixels