Dataset Versions

v38

2023-12-07 12:21pm

Generated on Dec 7, 2023

Dataset Split

Train Set 88%
182Images
Valid Set 9%
18Images
Test Set 4%
8Images

Preprocessing

Resize: Fit (black edges) in 1280x1280
Modify Classes: 25 remapped, 24 dropped
Filter Null: Require at least 99% of images to contain annotations.

Augmentations

Outputs per training example: 3
Crop: 0% Minimum Zoom, 20% Maximum Zoom
Rotation: Between -20° and +20°
Shear: ±20° Horizontal, ±20° Vertical
Hue: Between -50° and +50°
Saturation: Between -40% and +40%
Brightness: Between -40% and +40%
Exposure: Between -40% and +40%
Blur: Up to 1px
Noise: Up to 10% of pixels

Similar Projects

See More