Dataset Versions
Versions
v1
2023-11-22 9:29am
Generated on Nov 22, 2023
5486 Total Images
View All ImagesDataset Split
Train Set 92%
5051Images
Valid Set 5%
251Images
Test Set 3%
184Images
Preprocessing
Auto-Orient: Applied
Resize: Fit (black edges) in 1000x1000
Modify Classes: 5 remapped, 9 dropped
Filter Null: Require at least 90% of images to contain annotations.
Augmentations
Outputs per training example: 3
Flip: Horizontal
Rotation: Between -15° and +15°
Shear: ±15° Horizontal, ±15° Vertical
Saturation: Between -25% and +25%
Brightness: Between -25% and +25%
Noise: Up to 5% of pixels
Cutout: 8 boxes with 10% size each