inPHASE develops novel inference algorithms to recover quantitative phase from fringe-pattern images in techniques like interferometry, digital holography, moiré methods, and structured illumination. The project combines convolutional neural networks with dynamic nested sampling to decode phase information directly from recorded intensity distributions. It tackles the limits of classic methods and the poor generalization of some data-driven approaches by pursuing modality-agnostic, training-aware solutions. inPHASE is led at the Warsaw University of Technology and benefits from collaboration with the Max Planck Institute for Radio Astronomy and several international partners. The resulting algorithms are designed to deliver high-precision, non-contact measurements across nano- to macro-scale objects. Expected applications include biomedical cell analysis, optical component metrology, fluid-shape characterization, and experimental mechanics. 

Principal Investigator: Maria Cywińska, PhD

This section is currently being updated.