Dataset Versions

v5

2023-01-09 7:06am

Generated on Jan 9, 2023

Dataset Split

Train Set 87%
1695Images
Valid Set 8%
161Images
Test Set 4%
82Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 1024x512
Modify Classes: 2 remapped, 0 dropped

Augmentations

Outputs per training example: 3
Crop: 0% Minimum Zoom, 20% Maximum Zoom
Brightness: Between -35% and +35%
Exposure: Between -32% and +32%
Blur: Up to 1.75px
Noise: Up to 6% of pixels