Dataset Versions

2023-06-21 4:56pm

Generated on Jun 21, 2023

Train Set 88%

3246Images

Valid Set 7%

259Images

Test Set 5%

201Images

Auto-Orient: Applied

Resize: Fit (reflect edges) in 1280x1280

Auto-Adjust Contrast: Using Contrast Stretching

Modify Classes: 0 remapped, 1 dropped

Filter Null: Do not filter any null images.

Outputs per training example: 3

Rotation: Between -15° and +15°

Grayscale: Apply to 10% of images

Saturation: Between -10% and +10%

Exposure: Between -10% and +10%

Blur: Up to 1.5px

Bounding Box: Brightness: Between -10% and +10%

Bounding Box: Exposure: Between -10% and +10%

Bounding Box: Blur: Up to 2.5px

Bounding Box: Noise: Up to 5% of pixels