Sesame Disease Detection Using a Deep Convolutional Neural Network

Abstract

Sesame, along with coffee, is Ethiopia's most exported product and the country's main source of foreign exchange. It is mainly grown in the northern parts of Ethiopia. Sesame products varies from year to year due to various factors. One such factor is weather conditions and disease. There are different types of diseases affecting sesame crop and one of the major factors reducing the productivity of farmers. It is difficult to identify the neck eye disease species from large quantities of sesame farm. Having this, we proposed a step-by-step deep convolutional neural network to identify sesame disease. The proposed work follows six major image processing steps: image acquisition, image preprocessing, image segmentation, augmentation, feature extraction, and classification. The images were collected from the northern part of Ethiopia and Dejen wereda using the cameras of Samsung A32 phones and iPhone 6s with a resolution of 450 x 680 pixels. 540 images for Bacteria Blight infected, Phyllody infected and healthy classes were collected and fed to convolutional neural networks to extract key features from the segmented image and inserted into fully connected SoftMax layers to identify the respective sesame disease class. Our proposed model achieves 99% training accuracy, 98% testing accuracy.

Keywords

• deep learning
• convolutional neural network
• sesame disease