Dataset Versions

v4

2023-06-12 9:11pm

Generated on Jun 12, 2023

Dataset Split

Train Set 87%
4887Images
Valid Set 8%
441Images
Test Set 6%
315Images

Preprocessing

Resize: Stretch to 640x640
Auto-Adjust Contrast: Using Histogram Equalization
Tile: 3 rows x 3 columns

Augmentations

Outputs per training example: 3
Hue: Between -25° and +25°
Noise: Up to 1% of pixels
Mosaic: Applied
Bounding Box: Shear: ±21° Horizontal, ±15° Vertical
Bounding Box: Brightness: Between -17% and +17%
Bounding Box: Exposure: Between -29% and +29%
Bounding Box: Blur: Up to 3px