Semantic segmentation of radio maps with the Tiramisu model.
Train the model on your FITS data by running the train.py
script. The list of image paths needs to be specified by a text file where each line is a relative image path and the root data folder from which to look for files.
python train.py --dataset {PATH_TO_TXT_FILE} --data_dir {PATH_TO_DATA_DIR} --resume {PRETRAINED_WEIGHTS} --n_classes {N_CLASSES} --run-name {RUN_NAME}
To perform inference, run the streamlit file by running
python run app.py