The HIP Service Unit ‘Applied Computer Vision Lab’ supports Helmholtz researchers in the application of AI algorithms to their custom image analysis problems. Our services include designing tailored AI solutions, consulting in training data selection and (semi-automated) annotation, training of annotation workers, standardized evaluation of algorithm performance and the organization of image analysis competitions.
The HelpDesk is the central point of contact for all inquiries regarding the Helmholtz Imaging Platform and its services. Do not hesitate to contact us via helpdesk@helmholtz-imaging.de - we will either help you out directly or get you in contact with someone who can.
Identifying the best-fitting algorithm and adapting it to a specific image analysis problem is an involved process that requires profound expertise. Our service unit closely collaborates with domain experts with the goal of developing tailored solutions to your needs - just reach out to us, we are here to help!
Depending on the complexity of the task we also offer the opportunity of jointly applying for Helmholtz AI Projects.
Generating a high quality annotated dataset is a key requirement for designing and training successful AI algorithms. Generating such a database is, however, a major bottleneck in many research areas. Our service unit provides consulting in data selection, curation and annotation with the goal of accelerating the creation of large annotated datasets and avoiding common pitfalls.
We furthermore offer the application of semi-automated annotation methods to reduce the burden of purely manual annotation.
We develop methods that can generalize across different imaging modalities and domains. These algorithms will be make available as part of the HIP Solutions framework.
Evaluation of algorithms in a standardized environment is pivotal for enabling a fair comparison between methods. Such environments are, for example, international competitions. Our service unit develops and provides software for the standardized evaluation of both entire competitions and single algorithms.
From the perspective of a domain expert, competitions are a powerful platform to encourage a large number of AI experts to solve a specific image analysis problem. Should you have such a task at hand we can organize a competition aimed at solving it.