SIBI_LetterLevel_Dataset Dataset

v1

2024-06-17 2:08am

Generated on Jun 17, 2024

Dataset Split

Train Set 87%
20280Images
Valid Set 8%
1939Images
Test Set 4%
972Images

Preprocessing

Auto-Orient: Applied
Resize: Fill (with center crop) in 640x640
Filter Null: Require all images to contain annotations.

Augmentations

Outputs per training example: 3
Rotation: Between -15° and +15°
Shear: ±5° Horizontal, ±5° Vertical
Saturation: Between -15% and +15%
Brightness: Between -10% and +10%
Blur: Up to 0.5px
Noise: Up to 0.5% of pixels