Utilizing Aerial Imagery and Deep Learning Techniques for Identifying Banana Plants Diseases

¹ Quality Engineering Research Cluster, Malaysian Institute of Industrial Technology, Universiti Kuala Lumpur, 81700, Johor, Malaysia
² Center for Artificial Intelligence Technology, Universiti Kebangsaan Malaysia, Selangor, Malaysia
³ Faculty of Plantation and Agrotechnology, Universiti Teknologi MARA (UiTM), Cawangan Melaka Kampus Jasin, 77300 Merlimau, Melaka, Malaysia
⁴ UTM-MPRC Institute for Oil & Gas (IFOG), Universiti Teknologi Malaysia, Malaysia

^* Corresponding author: mohdaliff@unikl.edu.my

Abstract

The primary agricultural pursuit in Malaysia centres around banana cultivation; however, this vital crop faces the daunting challenge of multiple diseases that hinder its growth. The adverse consequences of these diseases extend beyond the farms to impact the nation’s economy. To empower farmers with the tools to promptly identify and categorize these diseases, image processing techniques offer a valuable solution. This research leverages deep learning Convolutional Neural Networks (CNN) implemented through MATLAB in conjunction with a DJI drone. By harnessing this technology, the system can automatically detect and classify major banana diseases. The study meticulously fine-tuned several hyperparameters to achieve impressive training and testing accuracy levels. The results revealed that the model attained its highest training accuracy of 81.27% at epoch 8 and its lowest accuracy of 78.40% at epoch 4, demonstrating its potential to aid in early disease detection and classification in banana crops.