Dataset Versions

v4

2024-01-29 6:32pm

Generated on Jan 29, 2024

Dataset Split

Train Set 88%
1083Images
Valid Set 8%
102Images
Test Set 4%
50Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x384
Filter Null: Require all images to contain annotations.

Augmentations

Outputs per training example: 3
Hue: Between -11° and +11°
Brightness: Between -7% and +7%
Exposure: Between -10% and +10%
Blur: Up to 1.8px
Noise: Up to 1.71% of pixels