README.md

An auto tool for sentinel2-L1C images downloading and processing(Cutting and removing unuseless dataset)

Follow the jupyter notebook for more details!

Note: imgread nad imgwrite functions in dataset_preprocess.py are implemented by @Neooolee

Step

Prepare your dataset dictionary like the following structure:

-- dataset
    -- S2A_MSIL1C_20190714T043711_N0208_R033_T46TFN_20190714T073938.SAFE   
    -- S2A_MSIL1C_20180930T030541_N0206_R075_T49QDD_20180930T060706.SAFE  

Create the imgs_list.txt to split the train and test likt the following structure:

"""
train
S2A_MSIL1C_20190714T043711_N0208_R033_T46TFN_20190714T073938

test
S2A_MSIL1C_20180930T030541_N0206_R075_T49QDD_20180930T060706

"""

Preprocess training dataset

python dataset_preprocess.py --source 'dataset' --output './prepared_dataset' --list 'imgs_list.txt'

Preprocess labels

python dataset_preprocess.py --source 'Mask' --output './prepared_dataset' --if_label True

Preprocess labels and organise them into the train/test split, removing any non-matching files

python dataset_preprocess.py --source 'Mask' --output './prepared_dataset' --if_label True --organise_labels True