API and Modules

linear_solvers

Classic (learning-free) linear assignment problem solvers.

classic_solvers

Classic (learning-free) two-graph matching solvers.

multi_graph_solvers

Classic (learning-free) multi-graph matching solvers.

neural_solvers

Neural network-based graph matching solvers.

utils

Utility functions: problem formulating, data processing, and beyond.

benchmark

The Benchmark module with a unified data interface to evaluate graph matching methods.

dataset

The implementations of data loading and data processing.

Warning

By default the API functions and modules run on numpy backend. You could set the default backend by calling pygm.set_backend('new_backend'). If you are using other backends, the corresponding package (such as PyTorch) should be installed. See the numerical backend guide for details.