Publicação
Vineyards monitoring using convolutional neural networks and multispectral images
| Resumo: | This study presents the development of an agricultural monitoring system designed to detect vineyards and crop lines through the application of binary segmentation techniques. The primary objective is to enhance the efficiency of vineyard monitoring, enabling precise plant detection using aerial imagery captured by unmanned aerial vehicles (UAVs). The system utilizes U-Net architecture for semantic segmentation, which was selected for its ability to effectively differentiate between vine and non-vine areas, promoting resource optimization and sustainable viticulture. Additionally, an algorithm based on the Hough Transform was implemented to accurately detect vineyard crop rows, further supporting precision agriculture practices. The model was trained and validated using datasets obtained from various sources, including publicly available datasets and those provided by industry partners. Evaluation metrics such as accuracy, Intersection over Union (IoU), and Dice Coefficient were employed to assess model performance, with results indicating varying levels of success across different datasets. The research contributes to the growing field of precision agriculture by offering a practical tool for vineyard management, with potential applications in resource allocation, environmental sustainability, and operational efficiency. The system’s design and the methodologies employed underscore the feasibility of integrating advanced machine learning models into real-world agricultural contexts.The code and dataset are publicly https://github.com/rodrigo-99ferreira/Vineyards |
|---|---|
| Autores principais: | Ferreira, Rodrigo Manso Teixeira Basílio |
| Assunto: | Agricultura de precisão Deteção de vinhas Segmentação binária Arquitetura U-Net Deteção de linhas de cultivo Teses de mestrado - 2024 |
| Ano: | 2024 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório da Universidade de Lisboa |
| Resumo: | This study presents the development of an agricultural monitoring system designed to detect vineyards and crop lines through the application of binary segmentation techniques. The primary objective is to enhance the efficiency of vineyard monitoring, enabling precise plant detection using aerial imagery captured by unmanned aerial vehicles (UAVs). The system utilizes U-Net architecture for semantic segmentation, which was selected for its ability to effectively differentiate between vine and non-vine areas, promoting resource optimization and sustainable viticulture. Additionally, an algorithm based on the Hough Transform was implemented to accurately detect vineyard crop rows, further supporting precision agriculture practices. The model was trained and validated using datasets obtained from various sources, including publicly available datasets and those provided by industry partners. Evaluation metrics such as accuracy, Intersection over Union (IoU), and Dice Coefficient were employed to assess model performance, with results indicating varying levels of success across different datasets. The research contributes to the growing field of precision agriculture by offering a practical tool for vineyard management, with potential applications in resource allocation, environmental sustainability, and operational efficiency. The system’s design and the methodologies employed underscore the feasibility of integrating advanced machine learning models into real-world agricultural contexts.The code and dataset are publicly https://github.com/rodrigo-99ferreira/Vineyards |
|---|