Dataset Versions
v2
RoboFlow Trainable
Generated on Dec 3, 2022
5398 Total Images
View All Images
Dataset Split
Train Set 92%
4983Images
Valid Set 6%
311Images
Test Set 2%
104Images
Preprocessing
Resize: Stretch to 640x640
Filter Null: Require all images to contain annotations.
Augmentations
Outputs per training example: 3
Flip: Horizontal
Crop: 0% Minimum Zoom, 50% Maximum Zoom
Rotation: Between -15° and +15°
Shear: ±15° Horizontal, ±15° Vertical
Hue: Between -30° and +30°
Saturation: Between -50% and +50%
Brightness: Between -35% and +35%
Exposure: Between -30% and +30%
Blur: Up to 3px
Noise: Up to 2% of pixels