Hyperautomation framework for real time vision-based quality control applications

This thesis describes the development and implementation of a hyperautomation framework based on a microservices architecture, specifically designed for industrial applications exploiting computer vision systems. The proposed solution is characterized by the easiness of use and fast implementation, providing a library of methods and functions to enable the exploitation of the most recent innovations in the deep learning field. The research was guided by a strong emphasis on real-world applications, resulting in a deep research focus not only on a software level, but also on the optimization of the hardware involved and the software/hardware implementation. This integrated approach allowed the realization of robust, reliable and customizable solutions, able to satisfy the strict constraints of timing and accuracy required for the modern industrial applications. In this sens, the thesis includes some real case studies demonstrating the effectiveness of the framework in different industrial scenario, highlighting the advantages of adopting this technology to improve the production process in terms of efficiency and precision.