Skaïgreen: An aerial drone and deep learning project for the counting and nitrogen deficiency detection of coconut trees

Abstract

This capstone project involved the implementation of a drone system to obtain an aerial view of a coconut farm, with the goal of using deep learning models to detect and count coconut plants, as well as identify nitrogen-deficient plants. The YOLOv5 algorithm was used for object detection, with a unique ID assigned to each plant for counting. A VGGNet convolutional neural network was trained on a dataset of labeled coconut plant images to classify plants as "healthy" or "unhealthy". The results of the project included an automated drone system that could fly and capture footage with a single click, allowing for efficient counting of coconut plants and detection of nitrogen-deficient plants.