Banca de QUALIFICAÇÃO: RENATO DE AVILA LOPES

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : RENATO DE AVILA LOPES
DATE: 09/02/2024
TIME: 09:00
LOCAL: https://conferenciaweb.rnp.br/sala/zampirolli
TITLE:

Computer vision for autonomous agricultural vehicles

PAGES: 70
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUBÁREA: Metodologia e Técnicas da Computação
SPECIALTY: Processamento Gráfico (Graphics)
SUMMARY:

Precision agriculture represents a technological revolution in which autonomous agricultural vehicles play a crucial role, aiming to increase efficiency and safety in monoculture cultivation. To achieve this goal, it is essential to implement a system that integrates a variety of sensors, with video cameras standing out as fundamental components. The purpose of this work lies in the application of image analysis through sensors installed on tractors or trucks, capable of identifying different types of plants, environments, and conditions. The most recent and efficient approaches to image classification involve the use of neural networks in visual analysis. Increasing operational safety requires applying machine learning processes to train neural networks to recognize and detect objects, people, and animals in the vehicle's operational area. The proposed method involves the development of a security system for autonomous vehicles based on computer vision and neural networks. This system employs machine learning processes to train neural networks to recognize and detect objects, people, and animals in the vehicle's operational area, contributing to making harvesting and transshipment operations safer and more efficient. Throughout the research, different pre-trained networks were compared, including Yolo 3, 4, 5, and 8, SSDMobilente V1 and V2, SSDInception V2, with Yolo 4 exhibiting the best performance when implemented on an AGX Jetson Orin board. This network achieved a frame rate of 5 FPS in the OpenCV library, demonstrating effectiveness in performing the proposed tasks.

COMMITTEE MEMBERS:
Presidente - Interno ao Programa - 1600876 - FRANCISCO DE ASSIS ZAMPIROLLI
Membro Titular - Examinador(a) Externo à Instituição - EDSON CAORU KITANI - FATEC-SA
Membro Titular - Examinador(a) Externo à Instituição - Leopoldo Rideki Yoshioka - USP
Membro Suplente - Examinador(a) Externo ao Programa - 2403225 - UGO IBUSUKI