pydegensac is a Python library for robust geometric estimation in computer vision, estimating homographies (H) and fundamental matrices (F) from sparse point correspondences using LO-RANSAC and DEGENSAC algorithms. The package exposes findHomography and findFundamentalMatrix through a pybind11 binding layer (src/pydegensac/bindings.cpp) backed by a C99 core (src/pydegensac/degensac/) implementing RANSAC sampling, MSAC scoring, degeneracy checking, local optimization, and LAF (local affine frame) consistency verification. A thin Python wrapper (src/pydegensac/utils.py) handles input validation, error-type mapping, and OpenCV KeyPoint conversion. Architecturally it is a hybrid: a legacy procedural C numerical core (including a vendored CCMATH-derived matrix utility library, matutls) compiled into static libraries, wrapped by pybind11, and shipped as a pip-installable extension module. It depends on LAPACK/BLAS for SVD and eigendecomposition and vendors pybind11 under lib/pybind11. The build is orchestrated by CMake with a custom setup.py CMakeBuild, and CI produces manylinux/macOS/Windows wheels via cibuildwheel for Python 3.8-3.14.