group B JCT

scanPray2

Classification

scanPray2 Image Dataset

v2

2023-01-23 12:18pm

Generated on Jan 23, 2023

Dataset Split

Train Set 87%
20862Images
Valid Set 8%
2001Images
Test Set 4%
1004Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640
Auto-Adjust Contrast: Using Adaptive Equalization

Augmentations

Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise
Crop: 0% Minimum Zoom, 60% Maximum Zoom
Rotation: Between -16° and +16°
Shear: ±15° Horizontal, ±15° Vertical
Brightness: Between -29% and +29%
Exposure: Between -25% and +25%
Blur: Up to 1.75px
Noise: Up to 7% of pixels