:snake:Highlight

SITSMamba adopts Mamba for Satellite image time series classification for the first time.

:ear_of_rice:Dataset

PASTIS
All the 20 classes are used for training and evaluation in our study.
MTLCC
All the 17 classes are used for training and evaluation in our study.
Please refer to GitHub - MarcCoru/MTLCC or GitHub - TUM-LMF/MTLCC-pytorch for instructions on loading the dataset and obtaining the eval.tileids.

:hammer:Environment

You can install mamba according to the following instruction.

conda create -n mamba python==3.8
pip install torch==2.2.1 torchvision==0.17.1 --index-url https://download.pytorch.org/whl/cu121
wget https://github.com/state-spaces/mamba/releases/download/v1.2.2/mamba_ssm-1.2.2+cu122torch2.2cxx11abiFALSE-cp38-cp38-linux_x86_64.whl
pip install mamba_ssm-1.2.2+cu122torch2.2cxx11abiFALSE-cp38-cp38-linux_x86_64.whl

:blue_heart:Thanks

The ConvBlock is from UTAE. The temporal encoder, Mamba block, is from mamba.