Dataset Versions

v6

Balanced Cracked and Uncracked- Cutouts Removed

Generated on Jun 19, 2024

Dataset Split

Train Set 92%
1025Images
Valid Set 5%
58Images
Test Set 3%
28Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640
Grayscale: Applied

Augmentations

Outputs per training example: 5
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Grayscale: Apply to 15% of images
Hue: Between -39° and +39°
Exposure: Between -10% and +10%
Blur: Up to 1.8px
Noise: Up to 10% of pixels
Bounding Box: Shear: ±10° Horizontal, ±10° Vertical

Similar Projects

See More