Dataset Versions

v18

Versao 4 - Valter

Generated on Dec 12, 2024

Dataset Split

Train Set 92%
1296Images
Valid Set 8%
116Images
Test Set %
0Images

Preprocessing

Auto-Orient: Applied
Static Crop: 26-75% Horizontal Region, 25-77% Vertical Region
Resize: Stretch to 640x640
Auto-Adjust Contrast: Using Adaptive Equalization
Tile: 2 rows x 2 columns

Augmentations

Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Rotation: Between -15° and +15°
Blur: Up to 2.5px
Noise: Up to 1.68% of pixels