Detection of Panama Disease on Banana Leaves Using the YOLOv4 Algorithm

Abstract

Manual visual analysis is the most used method in determining plant diseases. Although for many years, farmers and growers have depended on knowledge and experience in examining the condition of their plants for any possible ailments, which can be tedious and demanding. Bananas are the third most-sold crop in the Philippines. The Philippines is the top export of bananas within Asia, accumulating more than 80% of gross banana exports from Asian countries. Yet, for as long as the banana exportation industry has been running, banana plantations have constantly been suffering under the threat of various diseases emerging through the years. But the affliction that has significantly impacted the export trade in the Philippines is the Fusarium oxysporum, Panama Tropical Race 4, and Panama Wilt, more commonly known as the Panama Disease. By utilizing the YOLOv4 algorithm, we can save time and effort from manually analyzing and determining the possible presence of the disease, as well as achieve real-time judgment, thus, can potentially assist in reducing the tremendous loss caused by Fusarium oxysporum. The device developed has a 90.0% accuracy in determining the presence of Panama Disease through photos of banana leaves.