COVID-19 X-Ray datasets

COVID-19 annotated datasets created with open source images and annotated with bounding boxes, polygons and semantic segmentations of COVID-19 regions of interest. Metadata includes the patients age, sex and days since diagnosis as well as clinical notes and hospital information.

COVID-19 Machine learning image datasets

Our COVID-19 annotations are created by a multi-step human workflow involving Radiologists and talented image annotators. Our custom workflows, tools, and processes ensure the most accurate and cost effective annotations.


Experienced Radiologists manually segment anatomical and pathological features while talented annotators segment medical objects such as tubes, lines and probes.

Instance Segmentation

Radiologist segment the lungs in COVID-19 images which allows models to be trained on only the relevant parts of X-Ray images

Bounding Boxes

Talented annotators draw bounding boxes around the lungs which are checked by Radiologists for accuracy. Boxes can be upgraded to Polygons and Masks.