Machine learning (ML) has enabled and accelerated frontier research in the life sciences, but democratised access to such methods is, unfortunately, not a given. Access to necessary hardware and software, knowledge and training, is limited, while methods are typically insufficiently documented and hard to find. Furthermore, even though modern AI-based methods typically generalize well to unseen data, no standard exists to enable sharing and fine-tuning of pre-trained models between different analysis tools. In this seminar, we will discuss how the AI4Life consortium addresses such accessibility and standardisation gaps. We will present the unified DL model metadata specification and show the trained models we have collected in the BioImage Model Zoo. Importantly, the Model Zoo is supported by a large range of community partners, e.g. ilastik, Icy or BiaPy, that can all run models which adhere to the standard. This seminar is targeting users looking for pre-trained DL models to solve their bioimage analysis problems. We will also mention important features we have introduced for model developers such as our simple model contribution pipeline.
Speakers: Anna Kreshuk, Wei Ouyang