gbuzzard/PnP-MACE

Implements both classical Plug-and-Play priors using iterative denoiser integration and modern MACE (Consensus Equilibrium) solvers via Mann iterations, enabling flexible fusion of physics-based forward models with learned or algorithmic denoisers. Targets inverse problems in imaging by decoupling the reconstruction algorithm from the denoiser choice, allowing swappable deep learning or traditional approaches without retraining. Provides reproducible demos in Python/Jupyter with conda-based setup for immediate experimentation on subsampling and tomographic reconstruction tasks.

No commits in the last 6 months. Available on PyPI.