Convolutional neural networks and visual transformers applied to disease detection in plant leaf images

• Grant number: 23/02619-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date until: June 01, 2023
• End date until: May 31, 2024
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Nina Sumiko Tomita Hirata
• Grantee: Heitor Barroso Cavalcante
• Host Institution: Instituto de Matemática e Estatística (IME). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Associated research grant: 15/22308-2 - Intermediate representations in Computational Science for knowledge discovery, AP.TEM

Abstract

Techniques based on deep learning are currently the state of the artwhen it comes to processing various types of data like images, audio and text. In the Computer Vision field, convolutional neural networks and, more recently, visual transformers have promoted great advances in methods that perform information extraction from images. This research project aims to study theoretical and practical aspects of these two models. While the theoretical study will follow a traditional approach, the practical study will be carried out through training and assessment of networks on the problem of detecting diseases in images of plant leaves. The practical exercise will provide opportunities for comparing the two types of networks, including both quantitative and qualitative aspects. It is expected that the knowledge generated in this project can later be used to address a broader scope of the plant disease detection problem.