Dataset Versions

v2

2023-07-21 11:24pm

Generated on Jul 21, 2023

Dataset Split

Train Set 87%
1998Images
Valid Set 8%
189Images
Test Set 4%
99Images

Preprocessing

Static Crop: 25-94% Horizontal Region, 57-75% Vertical Region
Resize: Stretch to 512x512
Auto-Adjust Contrast: Using Histogram Equalization
Grayscale: Applied
Tile: 3 rows x 3 columns

Augmentations

Outputs per training example: 3
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Rotation: Between -15° and +15°
Bounding Box: Flip: Horizontal, Vertical
Bounding Box: Blur: Up to 12px
Bounding Box: Noise: Up to 5% of pixels