AMICAL

AMICAL: Assisted machine learning Model Improvement using Customer Active Learning

By exploiting our knowledge gained on previous projects (SoLiDA, ALDAAL and Bühler), we will improve our knowledge and competencies on machine learning algorithms regarding their training and validation processes. One very efficient way to improve the results of machine systems is to use an active learning methodology, that transfers human expertise to the machine models learned through data. We want to develop a generic methodology and we will validate our approach around two use cases: aerial images of vineyards and images of rice grain images, the evaluation of the use cases will be based on the comparison of the results of a baseline against multiple active learning steps.