Smart agriculture management system for rice disease detection and fertilizer
Abstract
Bangladesh has the ninth-highest population density in the world. Despite significant advancements to meet the expanding food-demand, the nation still relies heavily on agriculture, with unpredictable crop yields and insufficient infrastructure for farming as a result of many sectors adhering to outdated human-centric agricultural practices. A more sustainable agricultural-production could result from the use of smart-farming. Unmanned aerial vehicles (UAVs), a widely used smart-farming technology, are revolutionizing traditional farming methods. Knowing the capabilities of various systems enables the appropriate choice to be made in advance of a particular task, and each task necessitates the appropriate choice of the flight system. In this project, a UAV collected and uploaded images and soil-data from a paddy-field, including NPK, moisture, temperature, and humidity. On this dataset, we conducted an analysis to determine the field's fertility and give fertilizer recommendations. To identify rice diseases like brown spot, we also employed ResNet152V2 model, a deep learning image-based disease detection system. we have used our dataset (images taken with the UAV camera) for training and testing and we have used Google Colab for coding. Although the algorithm was fairly precise, the UAV model was quite unstable and the dataset was relatively tiny, necessitating a significant amount of work in this area in the future.