Orateur
Description
Adding novel processing stages to Area Detector normally requires developing a C++ plugin. This slows prototyping and limits access to modern analysis tools that are predominantly available in Python. We present a new Area Detector plugin that executes user-defined Python algorithms directly on acquired images and their metadata. The plugin provides access to advanced scientific libraries and machine learning toolkits, supports GPU acceleration, and enables parallel execution through multi-stage AD pipelines or Python multiprocessing. Algorithms can be updated without recompiling the IOC, allowing rapid iteration. An automatically generated GUI exposes configurable parameters to operators. The approach is being evaluated on ESA’s NEOSTED telescope system, where Python-based routines are used for telescope autofocusing.