Dataset Versions
v3
reflect image
Generated on Sep 9, 2023
6490 Total Images
View All Images
Dataset Split
Train Set 89%
5745Images
Valid Set 10%
665Images
Test Set 1%
80Images
Preprocessing
Auto-Orient: Applied
Resize: Stretch to 320x320
Augmentations
Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise
Rotation: Between -45° and +45°
Shear: ±24° Horizontal, ±24° Vertical
Hue: Between -14° and +14°
Saturation: Between -34% and +34%
Exposure: Between -10% and +10%
Blur: Up to 0.75px
Noise: Up to 2% of pixels
Cutout: 5 boxes with 10% size each