ANNOUNCEMENTS
This study aims to develop a comprehensive approach for maize cultivar classification and crop health assessment utilizing unmanned aerial vehicle (UAV) data and machine learning algorithms. The utilization of UAV data offers significant advantages in agricultural research and monitoring. The dataset has a resolution of 7cm in 5 multispectral bands at 475nm, 560nm, 668nm, 717nm and 840 nm, providing detailed and valuable information about crop health and cultivation patterns. That can be used to analyse and classify different maize cultivars, at different growth stages facilitating effective decision-making for farmers and agronomists in the region.
The primary objectives of this research are to create a classification model using machine learning algorithms and to develop a rudimentary crop health assessment model by selecting appropriate spectral vegetation indices and texture-based parameters. Constructing a data frame with a target attribute based on threshold values. Which will be used for binary crop health classification as “Good” or “Bad” using logistic regression. The initial results indicate an accuracy of 87% for the random forest model. And a result of 82% accuracy for logistic regression. The study aims to leverage these techniques to improve agricultural practices and support decision-making processes for farmers and agronomists in the region.
To further validate and refine the developed models, a larger temporal dataset will be collected and utilized. This expanded dataset will encompass multiple time points, allowing for a more comprehensive analysis and assessment of crop health over time. The models will be tested and validated using this extended dataset, providing a more accurate and robust evaluation of their performance.
Keywords: UAV, ML, Agriculture.
