Detección del Tizón Foliar en las hojas del cultivo de maíz (Zea Mays L.) mediante un modelo de visión por computador

Corn is one of the most cultivated and important cereals around the world, and it is affected by diseases such as leaf blight, generating a decrease in crop yield, causing economic losses in its production. This Curricular Integration Work (TIC) was to develop a computer vision model for the detection of leaf blight disease in the leaves of corn crops. The methodology used was based on the Cross-Industry Standard Process for the development of Machine Learning applications with Quality (CRISP-ML(Q)), adapting the following phases: data engineering, machine learning model engineering and learning model evaluation automatic. In the first phase, four sets of images were created, in the second phase the YOLOv7 model was adjusted for the detection of the disease together with the optimization of the hyperparameters and in the last phase a web prototype developed in Flask for the application was created. of the Zero-Shot Learning (ZSL) technique, obtaining a final accuracy of 97%. In addition, the web prototype was evaluated by applying a survey to those involved in the Production Systems subject of the Agronomy Career at the National University of Loja, reaching an average of 96% in the Perceived Utility (PU) variable. and 97.2% in the Perceived Ease of Use (PEU), determining great usefulness and use of the web prototype that uses the computer vision model for the detection of leaf blight in corn leaves, being a support tool for the matter. Keywords: object detection, CRISP-ML(Q), YOLOv7, Zero-shot learning.

• Autor Principal: Jaramillo Cárdenas, Jennifer Jazmín (author)
• Formato: bachelorThesis
• Idioma: spa
• Publicado: 2024
• Subjects: DETECCIÓN DE OBJETOS; CRISP-ML(Q); YOLOV7; ZERO-SHOT LEARNING
• Acceso en liña: https://dspace.unl.edu.ec/jspui/handle/123456789/29459