Dataset Versions

v3

2024-11-19 12:06pm

Generated on Nov 19, 2024

Dataset Split

Train Set 88%
5412Images
Valid Set 8%
512Images
Test Set 4%
256Images

Preprocessing

Auto-Orient: Applied
Tile: 2 rows x 2 columns
Modify Classes: 2 remapped, 0 dropped

Augmentations

Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise
Noise: Up to 0.1% of pixels