Deep Graph Matching library with PaddlePaddle

In this open-source project, I engineered the PaddlePaddle version of a Graph matching model library, a.k.a., ThinkMatch, including SOTA Deep GM models and benchmarks. I implemented whole deep learning pipeline (dataset, model, training procedure) in Paddle and a PyTorch-to-Paddle pretrained-weights conversion tool, and donated bug findings to PaddlePaddle.

The whole project has received over 700 stars in Github, and my code also serves as a reference code in the course AI3607 Deep Learning and its Application at Shanghai Jiao Tong University.