Pytorch implementation of DeMatch++ for a comprehensive extension of CVPR'24 paper "DeMatch: Deep Decomposition of Motion Field for Two-View Correspondence Learning".
This repo contains the code and data for essential matrix estimation.
Please use Python 3.6, opencv-contrib-python (3.4.0.12) and Pytorch (>= 1.7.0). Other dependencies should be easily installed through pip or conda.
For a quick start, clone the repo and download the pretrained model.
git clone https://github.com/SuhZhang/DeMatchPlus
cd DeMatchPlus Then download the pretrained models from here.
Then run the feature matching with demo.
cd ./demo && python demo.pyWe provide the model trained on YFCC100M and SUN3D. Run the test script to get results (the generated putative matches can be downloaded from here).
cd ./test
python test.pyYou can change the default settings for test in ./test/config.py.