Dataset Versions

v4

2024-08-12 4:25am

Generated on Aug 12, 2024

Dataset Split

Train Set 87%
516Images
Valid Set 8%
49Images
Test Set 4%
25Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640

Augmentations

Outputs per training example: 3
Flip: Horizontal
Crop: 0% Minimum Zoom, 26% Maximum Zoom
Grayscale: Apply to 15% of images
Exposure: Between -27% and +27%
Blur: Up to 1.8px
Noise: Up to 0.7% of pixels
Cutout: 9 boxes with 6% size each