Dataset Versions

v6

2023-09-27 12:36pm

Generated on Sep 27, 2023

Dataset Split

Train Set 90%
3900Images
Valid Set 7%
296Images
Test Set 4%
156Images

Preprocessing

Auto-Orient: Applied
Static Crop: 41-67% Horizontal Region, 23-93% Vertical Region
Resize: Stretch to 640x640
Tile: 2 rows x 2 columns
Modify Classes: 2 remapped, 0 dropped

Augmentations

Outputs per training example: 3
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Cutout: 4 boxes with 26% size each
Mosaic: Applied
Bounding Box: Exposure: Between -12% and +12%