Code for Vision transformers in burned area delineation by Daniele Rege Cambrin, Luca Colomba and Paolo Garza. The code will be released once ready for use. Some parts may be missing.

To a full list of dependency look at requirements.txt. They can be installed with pip install -r requirements.txt.

Simply run main.py with arguments:

• ckpt_path: path of the checkpoint to load (default: None)
• steps: list of steps to execute (default: train)
• tag: comet experiment tag (default: test)
• config_file: configuration file path (default: configs/reduced_config.yaml)
• seed: for reproducibility (default: 7)
• test_set: the name of the fold used as test set. (default: purple)

This project is licensed under the Apache 2.0 license. See LICENSE for more information.

This repository use code from other repository. See their licenses in licenses folder.

We have to thanks HPC@POLITO and SmartData@POLITO for the provided computational resources.