Guideline for the development of deep learning image recognition systems

Abstract

This DIN SPEC in accordance with the PAS procedure has been drawn up by a DIN SPEC (PAS)-consortium set up on a temporary basis. This DIN SPEC (PAS) has been developed and approved by the authors named in the foreword. This DIN SPEC (PAS) defines requirements for the development of deep learning image recognition systems. This document gives the conditions, under which image recognition problems can be processed with the help of a Deep-Learning-System. It allows decision makers to gain knowledge about the application possibilities of a Deep-Learning-System and its structure. With the help of this document an estimation about the outlay and use of a Deep-Learning-System can be supported, and a more precise success prognosis can be made. This document gives guidelines for the practical implementation of a Deep-Learning-System starting with processes for data collection over the right structuring of data for AI-image recognition learning to the operational structure of learning experiments and error analyses. This document is meant for decision makers of AI-projects for image recognition as well as for those implementing such projects. This document does not address specific requirements for the fields of active learning, meta learning and continuous learning. © 2022 Beuth Verlag GmbH

External Links