KLASIFIKASI POHON KELAPA SAWIT PADA DATA FUSI CITRA LIDAR DAN FOTO UDARA MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORK

Abstract

Oil palm tree growth monitoring process is one of the important aspects that determine the quality of oil palm production. This process takes a long time and is very difficult for humans to do. The future of oil palm will be more prospective due to rapid technological advances and increasing human awareness of environmental sustainability. Remote sensing technology is currently being developed in the fields of plantations and agriculture, one of which is aerial photography and LiDAR technology. However, in Indonesia the use of LiDAR technology for mapping is not very popular because it is a new technology and many plantation communities are not familiar with the technology. Because this technology is relatively new and many plantation communities are not familiar with the technology. The focus of this research is the initial stage in developing a remote monitoring system for oil palm trees using LiDAR data and aerial photographs, namely the classification stage. Work was carried out starting from the data collection process and experiments to obtain an optimal classification model for the identification of oil palm trees. The data used are LiDAR image and aerial image fusion data of oil palm plantation areas in Pontianak, West Kalimantan, Indonesia. The classification method used is Convolutional Neural Network, with the highest accuracy using the RGB feature of 98%, the lowest accuracy using the LiDAR feature of 86%, while the combination of LiDAR fusion data and aerial photography is 97%.