Dataset Versions

v7

2024-08-16 4:54pm

Generated on Aug 16, 2024

Dataset Split

Train Set 89%
342Images
Valid Set 3%
10Images
Test Set 8%
31Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640
Filter Null: Require all images to contain annotations.

Augmentations

Outputs per training example: 3
Flip: Horizontal
Cutout: 6 boxes with 15% size each
Bounding Box: Rotation: Between -24° and +24°
Bounding Box: Brightness: Between -18% and +18%
Bounding Box: Blur: Up to 1.3px