Dataset Versions

v8

dataaugmentation

Generated on Apr 1, 2023

Popular Download Formats

Pascal VOC XML
Common XML annotation format for local data munging (pioneered by ImageNet).
PaliGemma
PaliGemma JSONL format used for fine-tuning PaliGemma, Google's open multimodal vision model.
CreateML JSON
CreateML JSON format is used with Apple's CreateML and Turi Create tools.
Other Formats
Choose another format.

Dataset Split

Train Set 87%
552Images
Valid Set 8%
48Images
Test Set 5%
32Images

Preprocessing

Auto-Orient: Applied
Isolate Objects: Applied
Static Crop: 41-77% Horizontal Region, 38-51% Vertical Region
Resize: Stretch to 640x640
Tile: 2 rows x 4 columns

Augmentations

Outputs per training example: 3
Flip: Horizontal, Vertical
90° Rotate: Clockwise, Counter-Clockwise, Upside Down
Crop: 0% Minimum Zoom, 99% Maximum Zoom
Rotation: Between -45° and +45°
Shear: ±31° Horizontal, ±45° Vertical
Grayscale: Apply to 88% of images
Hue: Between -101° and +101°
Saturation: Between -99% and +99%
Brightness: Between -30% and +30%
Exposure: Between -21% and +21%
Blur: Up to 25px
Noise: Up to 14% of pixels
Cutout: 18 boxes with 91% size each
Mosaic: Applied
Bounding Box: Flip: Horizontal, Vertical
Bounding Box: 90° Rotate: Clockwise, Counter-Clockwise
Bounding Box: Crop: 0% Minimum Zoom, 99% Maximum Zoom
Bounding Box: Rotation: Between -39° and +39°
Bounding Box: Shear: ±30° Horizontal, ±28° Vertical
Bounding Box: Brightness: Between -56% and +56%
Bounding Box: Exposure: Between -70% and +70%
Bounding Box: Blur: Up to 25px
Bounding Box: Noise: Up to 25% of pixels