This repository is based on DeepHistReg framework and uses LoFTR with QuadTree Attention for affine registration. We recommend to use this version as it fixes a bug related to compose_displacement_field function (which resulted in artifacts near image borders) and adopts standart STN implementation

Clone repo and cd:

git clone https://github.com/VladPyatov/ImgRegWithTransformers.git

cd ImgRegWithTransformers

Create and activate virtual environment:

python3 -m venv env

source env/bin/activate

Configure DeepHistReg framework as suggested by authors

Install LoFTR dependencies from requirements.txt:

python3 -m pip install --upgrade pip

pip install -r networks/feature_matching/requirements.txt

Install QuadTree Attention module:

cd ./networks/QuadTreeAttention && python setup.py install
cd ../../

Configure main.py and run registration:

python3 main.py