Dataset Versions

v3

avec augmentation

Generated on Dec 4, 2024

Dataset Split

Train Set 88%
246Images
Valid Set 8%
23Images
Test Set 4%
12Images

Preprocessing

Resize: Stretch to 640x640

Augmentations

Outputs per training example: 3
Rotation: Between -5° and +5°
Shear: ±10° Horizontal, ±10° Vertical
Exposure: Between -10% and +10%
Blur: Up to 0.7px
Noise: Up to 1.01% of pixels