Download the dataset here, and unzip the dataset to polyps/data. We use a similar dataset as PraNet, with improved pre-processing and filtering of duplicates. See polyps/process_data.ipynb for details ...