Dataset Versions

v4

2023-11-13 3:05pm

Generated on Nov 13, 2023

Dataset Split

Train Set 88%
2028Images
Valid Set 8%
192Images
Test Set 4%
96Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640
Tile: 2 rows x 2 columns
Filter Null: Require at least 40% of images to contain annotations.

Augmentations

Outputs per training example: 3
Rotation: Between -30° and +30°
Shear: ±14° Horizontal, ±19° Vertical
Exposure: Between -25% and +25%
Blur: Up to 2.75px