Hybrid Sample Generation

This project extracts real anomalies from labelled 2D images or 3D volumes, trains a generative model, creates multiple synthetic variants and places them into original control samples. It is based on the IEEE paper AMLDS63918.2025.11159383.

Start here

• Install the project and run a study.
• Learn about input data, pipeline stages, and resuming a study.
• Browse the configuration reference or use the search box to find a parameter.
• Read about study storage, evaluation, and visualization.
• See citation and license.