Dataset Versions
v6
2023-06-20 8:15pm
Generated on Jun 20, 2023
3735 Total Images
View All ImagesDataset Split
Train Set 88%
3273Images
Valid Set 7%
261Images
Test Set 5%
201Images
Preprocessing
Auto-Orient: Applied
Resize: Fit (white edges) in 1280x1280
Auto-Adjust Contrast: Using Histogram Equalization
Filter Null: Do not filter any null images.
Augmentations
Outputs per training example: 3
Rotation: Between -15° and +15°
Grayscale: Apply to 15% of images
Saturation: Between -10% and +10%
Exposure: Between -10% and +10%