facerecognition

face

Classification

face Image Dataset

v1

2024-04-14 11:42pm

Generated on Apr 14, 2024

Dataset Split

Train Set 88%
7620Images
Valid Set 8%
724Images
Test Set 4%
362Images

Preprocessing

Auto-Orient: Applied
Resize: Stretch to 640x640

Augmentations

Outputs per training example: 3
Crop: 0% Minimum Zoom, 10% Maximum Zoom
Rotation: Between -19° and +19°
Shear: ±10° Horizontal, ±10° Vertical
Saturation: Between -25% and +25%
Brightness: Between -40% and +40%
Exposure: Between -10% and +10%
Blur: Up to 2.5px
Noise: Up to 2.71% of pixels
Cutout: 5 boxes with 10% size each