Dataset Versions

v7

2024-05-01 3:28pm

Generated on May 1, 2024

Dataset Split

Train Set 87%
726Images
Valid Set 9%
71Images
Test Set 4%
33Images

Preprocessing

Static Crop: 3-97% Horizontal Region, 11-88% Vertical Region
Resize: Fill (with center crop) in 640x640
Auto-Adjust Contrast: Using Contrast Stretching
Filter Null: Require at least 65% of images to contain annotations.

Augmentations

Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Rotation: Between -25° and +25°
Shear: ±20° Horizontal, ±20° Vertical
Saturation: Between -30% and +30%
Brightness: Between -31% and +31%
Blur: Up to 1.5px
Noise: Up to 0.33% of pixels