Dataset Versions

v11

2024-10-16 2:39am

Generated on Oct 16, 2024

Dataset Split

Train Set 88%
10794Images
Valid Set 8%
989Images
Test Set 4%
500Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 1280x1280
Modify Classes: 39 remapped, 4 dropped
Filter Null: Require all images to contain annotations.

Augmentations

Outputs per training example: 3
Hue: Between -15° and +15°
Exposure: Between -10% and +10%
Blur: Up to 2.5px
Noise: Up to 1.01% of pixels