27-10-2023 ;11:30 DOCTORAL THESIS DEFENCE MIKEL LABAYEN ESNAOLA
First publication date: 19/10/2023
![](https://www.ehu.eus/documents/340468/13175559/tesis+doctoral+PANTALLAS+Mikel+Labayen+.jpg/9c0325bd-b930-63da-17dc-db77dd9b42b4?t=1697723054794)
Mikel Labayen Esnaola: “Técnicas de Visión e Inteligencia Artificial para Aplicaciones Industriales en Producción “.
Zuzendariak_Directores: Basilio Sierra Araujo/Naiara Aginako Bengoa.
2023_10_27, 11:30 Gela / Aula 3.2
Abstract:
"Computer Vision (CV) is very present both in everyday personal applications used in smartphones, computers or even vehicles, and in the production industry, where it is one of the foundational technologies for the automation of many visual processes (automated manufacturing and control, audiovisual content analysis, video surveillance, autonomous driving ... ). This research line (CV), which belongs to the so-called Artificial Intelligence (Al), has helped to add value and improve the quality of products, as well as managing manufacturing and facilitating the marketing and distribution of goods. In general, it has enabled the creation of new value-added functionalities and the optimisation of numerous processes, decreasing their costs and making them more efficient. Machine vision began its journey in industry at the end of the first decade of the 21st century (2005-2012). At that time, classical image processing techniques faced the challenge of reducing complexity in their SW algorithms and making them run in real time on an acceptable HW set-up in terms of cost and size. With the Deep Learning (DL) revolution that started in the second decade of this century and has developed especially in the last few years (2012-present), CV-based systems have been endowed with Al, introducing the use of techniques based on Deep Neural Networks (DNN). These neural networks have quickly become an essential part of the CV field, given their potential and applicability, demonstrated in many industry sectors. However, the introduction of DNNs in industrial products/services/applications brings with it a number of challenges that need to be solved gradually. Today, sorne of them are still awaiting a real and feasible solution. The choice of SW and HW architectures, as well as their configuration/modification for specific use cases, require new research lines and tasks in arder to guarantee the Reliability, Availability, Maintainability, Scalability (RAMS), and ultimately, if the functionality requires it, the Security of the products/services so that systems based on these technologies can be marketed. This thesis project aims to provide knowledge and tools based on Computer Vision and Artificial Intelligence ( CV&AI) for the development of:
■ Functional and accessible prototypes for any type of company.
■ Commercial products in production.
■ Products for embedded systems.
■ Safety lntegrity Level (SIL) type functionalities.
This work also presents the results of research carried out in different use cases and in different industrial sectors throughout the author's professional life. This research has been published in international journals and conferences, in order to obtain the support of the people who have accepted to publish the articles, and of the people who are willing to read them.
This report, which is a compilation of articles, presents a broad summary of the research work carried out, as well as the articles and patents that support it. "