Weed identification technique in basil crops using computer vision

• Autores: Yauri Rodríguez, Ricardo, Guzman Rojas, Brayan Joel, Hinostroza Gonzales, Alan, Gamero, Vanessa
• Formato: artículo
• Fecha de Publicación: 2023
• Institución: Universidad Tecnológica del Perú
• Repositorio: UTP-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorio.utp.edu.pe:20.500.12867/7803
• Enlace del recurso: https://hdl.handle.net/20.500.12867/7803; http://doi.org/10.37394/23202.2023.22.64
• Nivel de acceso: acceso abierto
• Materia: Agriculture; Crops; Image processing; Computer vision; https://purl.org/pe-repo/ocde/ford#1.02.00; https://purl.org/pe-repo/ocde/ford#4.01.01

The promotion of organic and ecological production seeks the sustainable and competitive growth of organic crops in countries like Peru. In this context, agro-exportation is characterized by-products such as fruit and vegetables where they need to comply with organic certification regulations to enter products into countries like the US, where it is necessary to certify that weed control is carried out using biodegradable materials, flames, heat, media electric or manual weeding, this being a problem for some productive organizations. The problem is related to the need to differentiate between the crop and the weed as described above, by having image recognition technology tools with Deep Learning. Therefore, the objective of this article is to demonstrate how an artificial intelligence model based on computer vision can contribute to the identification of weeds in basil plots. An iterative and incremental development methodology is used to build the system. In addition, this is complemented by a Cross Industry Standard Process for Data Mining methodology for the evaluation of computer vision models using tools such as YOLO and Python language for weed identification in basil crops. As a result of the work, various Artificial Intelligence algorithms based on neural networks have been identified considering the use of the YOLO tool, where the trained models have shown an efficiency of 69.70%, with 3 hours of training, observing that, if used longer training time, the neural network will get better results