Portable Sigatoka Spot Disease Identifier on Banana Leaves Using Support Vector Machine
Giorgette Louise H. Tuazon, Hazeline M. Duran, Jocelyn F. Villaverde
Abstract
Banana plant diseases serve as a threat to farmers which can affect the quality and quantity of banana production. Black Sigatoka Spot is considered as the most economically important banana leaf disease as it affects banana plantains worldwide, causing yield loses up to 50%. It is caused by a fungi called Mycosphaerella fijensis. Sigatoka spots are small reddish-brown flecks that are seen on the leaf and overtime develop into pale yellow, brown streaks surrounded by a yellow halo. The goal of this research is to develop a portable system that can identify the sigatoka spot disease on a banana leaf to help famers accurately identify Sigatoka spot on banana leaves. The output of the system displays the classification of the leaf and the leaf area. The researchers used a USB Camera to capture banana leaf images and were processed by stitching, equalization, segmentation, feature extraction and classification by Support Vector Machine. Overall, the proposed system obtained a 90% accuracy in classifying healthy and sigatoka-infected banana leaves.