Fig. 1. Overview of competition setup.

Tissue data for five organs collected within HPA and HuBMAP using different tissue harvesting and processing protocols are collected and divided into a public training, public test, and private test dataset. Kaggle teams use the public training data to train their models; they then submit the models to the Kaggle Submission Portal to receive performance scores computed using the public test data. At the end of the competition, when all teams submit their best algorithm solutions, their score on the private test set determines the Performance Prize winners.