Installation⚙️

Requirement:

• Linux with Python ≥ 3.10.
• PyTorch ≥ 2.0 and torchvision that matches the PyTorch installation.

Set up conda envirnment:

conda create -n segearthr2 python=3.10
conda activate segearthr2

git clone https://github.com/earth-insights/SegEarth-R2.git
cd segearth_r2

pip install torch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 --index-url https://download.pytorch.org/whl/cu121

pip install -r requirements.txt

Install detectron2

Follow detectron2

CUDA kernel for MSDeformAttn

After preparing the required environment, run the following command to compile CUDA kernel for MSDeformAttn:

cd segearth_r2/model/mask_decoder/Mask2Former_Simplify/modeling/pixel_decoder/ops
sh make.sh

Note

If you encounter any difficulties during the environment configuration process, you can directly download the pre-compiled environment package here

tar -xzvf segearthr2.tar.gz
source segearthr2/bin/activate