Abdari, Ali (2026) Educational Agriculture in Metaverse. [Tesi di dottorato]

Documento PDF Abdari_Ali_38.pdf Visibile a [TBR] Amministratori dell'archivio Download (7MB) | Richiedi una copia

Abstract

In recent years, we have witnessed many problems such as economic issues, population growth, and most importantly climate changes, which have affected different aspects of our lives. One of the sectors that has been highly influenced is food production, and more generally, the agricultural sector. Therefore, there is a strong need for farmers to learn efficient and effective farming techniques to produce high-quality products at reasonable prices. However, access to suitable educational content is not always easy. In addition, relying on experts can be expensive and, in many cases, not practical, especially for small farmers. At the same time, with the rapid growth of technology, particularly artificial intelligence and virtual environments such as VR and AR, the concept of the metaverse has gained significant attention. Using new ideas like the metaverse can provide accessible learning environments where many people can learn new skills in a cost-effective way. For instance, a user may find a metaverse presenting an interactive environment to learn the best practices in pruning fruit trees, merging “theory” (e.g., videos describing key notions) and “practice” (e.g., a virtual experience to explore pruning with guidance). Since these environments can be attractive and interactive, the learning process can become more interesting and easier for users. With these experiences becoming increasingly available to the wider public, users find it difficult to identify those they are most interested in among many irrelevant ones. This problem has not been properly addressed so far in the literature. In this thesis, the task of retrieving complex 3D scenes using natural language is studied, with a focus on its application in educational agriculture. To address this problem, different aspects of the task are explored, including the creation of suitable datasets, methods for representing 3D scenes, techniques for processing natural language, and effective ways of training cross-modal systems. Finally, the application of these methods in educational agricultural environments is investigated, showing their potential to support learning in a more interactive and accessible way.

Downloads

Actions (login required)

Modifica documento