Dataset Versions

v3

2023-05-31 12:19am

Generated on May 30, 2023

Dataset Split

Train Set 92%
7221Images
Valid Set 8%
604Images
Test Set %
0Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640
Modify Classes: 15 remapped, 6 dropped

Augmentations

Outputs per training example: 3
Rotation: Between -32° and +32°
Shear: ±43° Horizontal, ±40° Vertical
Grayscale: Apply to 26% of images
Hue: Between -108° and +108°
Saturation: Between -67% and +67%
Brightness: Between -25% and +25%
Blur: Up to 2.5px
Noise: Up to 5% of pixels
Mosaic: Applied
Bounding Box: Rotation: Between -37° and +37°
Bounding Box: Shear: ±36° Horizontal, ±36° Vertical
Bounding Box: Blur: Up to 2.5px
Bounding Box: Noise: Up to 5% of pixels