Stem-base Rot Disease Detection in Oil Palm using RGB (Red, Green, Blue) and OCN (Orange, Cyan, NIR) Image Fusion Method Based on ResNet50

Current image acquisition and processing methods still need to be improved to effectively detect oil palm diseases. A precise and fast method to detect stem base rot disease in oil palm trees can be developed using drone technology and image processing approaches. An OCN (Orange, Cyan, NIR) camera is added to a standard drone and equipped with an RGB (Red, Green, Blue) camera. Combining the two cameras is proposed to generate multispectral imagery using an image fusion method called early fusion. A Multispectral Convolution Neural Network (MCNN) is also introduced to detect stem base rot disease by analysing the leaf patterns of oil palms. Healthy and unhealthy leaf samples were collected from oil palm plantations in Bogor. The images that have passed the image processing stage with the fusion method become inputs for modelling to identify stem base rot disease in oil palm. The results of the research using the multispectral image fusion method (RGB and OCN) based on the ResNet50 architecture can be used to identify stem base rot disease in oil palm effectively, as evidenced by the training and validation accuracy of 97.75% and 96.48%.