--- proposal-title: ML training on PaN open data challenge: cross-RI domain: 3. Photon/neutron sources-based experimental research consortium: ALBA duration: 2 years costs: 250 k€ description: > To apply efficient deep learning crystal segmentation approach to a new crystallography technique, this project will collect, structure, curate and publish experimental data of segmented crystals. impacts: > FT-SSX is an emerging technique in photon facilities, this work would benefit the global community currently building around it. It would also serve as a blueprint for other ML applications to be built around open data. resources: - roadmap for ML-ready open data in PaN - metadata framework - federated PaN search API - PaN open data search portal - PaN training catalogue ---