Dataset Analytics
Generated on

Number of Images

missing annotations
null examples

Number of Annotations

per image (average)
</> Across classes

Average Image Size

mp
🔍 from
🔍 to

Median Image Ratio

Classes i

Overview of the number of annotations for each class in your dataset.
Train Valid Test

Diagnosis Distribution i

HAM10000 lesion type recovered from each filename prefix — the label the stage-2 classifier predicts.

Dimension Insights i

Sizes and aspect ratios of the images in your dataset. Dashed lines are the median width ( px) and median height ( px).
Images

Annotation Insights i

How large the boxes are, how they are shaped, and where they sit inside the frame.
Box size (√area as % of image)
Aspect ratio (width ÷ height)
Annotation heatmap
Annotations per image

Dataset Split i

Train / valid / test partition as it exists on disk.