Browsing by Author "Torres Cardozo, Maicol Stiven"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
Item Metodología para clasificar el nivel de estrés utilizando imágenes hiperespectrales en cultivo de aguacate(Universidad Santiago de Cali, 2023) Carrera Carabali, Sebastian; Torres Cardozo, Maicol Stiven; Sarria-Paja, Milton Orlando (Director)This article focuses on a feasibility study where the application of remote sensing based on hyperspectral images is used to classify the level of water stress in avocado crops using artificial intelligence techniques. This exploration involves employing various preprocessing methods based on representative bands of hyperspectral images. It’s important to highlight that hyperspectral images are highly informative data sources, necessitating the use of different techniques for effective data management, enabling the analysis of imperceptible data. Therefore, the process of label generation and partitioning the database into pixel neighborhood information and localized information is detailed. Additionally, the use of vegetation indices in the agricultural analysis of avocado crops is essential for identifying areas of interest. Furthermore, it provides a detailed insight into data analysis, statistics, and the performance of classification models in the context of spectral reflectance of avocado genotypes, pinpointing challenges and areas for improvement that enhance the efficiency and precision of avocado crop monitoring, reducing the time associated with manual evaluation and providing a tool for farmers. Thus, the prediction results vary for the two avocado genotypes analyzed using convolutional and fully connected networks. This strategy achieves accuracy percentages ranging from 4.4% to 42.89%, estimating crucial factors to consider in the analyses.