Unmanned Aerial Vehicle for Remote Sensing Detection of Oil Palm Trees Using You Only Look Once and Convolutional Neural Network

Oil palm (Elaeis Guineensis Jacq) is one of several palm species that produces oil for commercial purposes. Indonesia is the largest producer of palm oil in the world, thanks to its extensive plantations. The marketing of palm oil and palm kernel in Indonesia has great potential both domestically and internationally. However, the management of oil palm plantations, including the counting of palm trees, poses challenges due to the vastness of the plantations. Conventional methods of counting palm trees from aerial imagery are time-consuming and require significant additional costs. This research aims to address these issues and contribute to modern agriculture or precision farming. The dataset used consists of 500 unmanned aerial vehicle remote sensing images of oil palm trees, with an 80% training data split, 10% testing data, and 10% validation data. The algorithm used is You Only Look Once Version 7 with a convolutional neural network architecture. Testing was performed on 50 testing data of oil palm trees. The results of this study indicate that the model's performance in detecting oil palm trees achieved a precision rate of 98.45%, recall rate of 98.17%, overall accuracy of 98.31%, and a mean average precision of 99.7%.