Establishing the correct protocol for treating COVID-19 and estimating the specific percentage of COVID-19 infection in lung tissue can be an important tool.

This article describes the approach we used to approximate the percentage of COVID-19 on lung CT scan sections from the Covid-19-Infection-Percentage-Estimation-Challenge.

Our method is based on modern training pipelines and architectures used to train state-of-the-art models on image classification tasks.

We achieved the best score on the validation dataset for the Covid-19 Infection Percentage Estimation competition.

The final model consists of a set of 40 models and achieved a score of 4.17140661 MAE, 0.948787386 PC, and 8.196144648 RMSE.