Dataset Versions
Versions
v1
2024-09-19 9:35am
Generated on Sep 19, 2024
38328 Total Images
View All Images
Dataset Split
Train Set 88%
33540Images
Valid Set 8%
3184Images
Test Set 4%
1604Images
Preprocessing
Auto-Orient: Applied
Static Crop: 24-74% Horizontal Region, 25-75% Vertical Region
Tile: 2 rows x 2 columns
Augmentations
Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Rotation: Between -20° and +20°
Brightness: Between -35% and +35%
Exposure: Between -12% and +12%
Noise: Up to 7.15% of pixels
Cutout: 13 boxes with 13% size each