Dataset Versions

v1

2023-12-04 3:02pm

Generated on Dec 4, 2023

Dataset Split

Train Set 93%
2619Images
Valid Set 5%
144Images
Test Set 2%
54Images

Preprocessing

Auto-Orient: Applied
Isolate Objects: Applied
Resize: Stretch to 640x640
Auto-Adjust Contrast: Using Adaptive Equalization
Tile: 3 rows x 3 columns

Augmentations

Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Rotation: Between -15° and +15°
Brightness: Between -40% and +40%
Blur: Up to 2.5px