Dataset Versions

v1

2023-07-04 3:59pm

Generated on Jul 4, 2023

Dataset Split

Train Set 87%
111Images
Valid Set 8%
10Images
Test Set 5%
6Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640

Augmentations

Outputs per training example: 3
Crop: 0% Minimum Zoom, 57% Maximum Zoom
Hue: Between -31° and +31°
Brightness: Between -58% and +58%
Exposure: Between -30% and +30%
Blur: Up to 2.5px
Noise: Up to 13% of pixels
Bounding Box: Flip: Horizontal, Vertical