A CNN Approach for Corn Leaves Disease Detection to support Digital Agricultural System
Kshyanaprava Panda Panigrahi, Abhaya Kumar Sahoo, Himansu Das
Abstract
Correct, fast and early detection of corn leave diseases and their prevention and control at the earlier stages is profitable. To improve the detection accuracy of corn leaf diseases, a CNN model has been approached. The improved CNN model is used for training and testing four kinds of corn leaf images are achieved by adding rectified linear unit activation functions and adam optimizer, by adjusting the parameters, pooling operations and reducing the number of classifiers. During the detection of three types of corn leaf diseases, this model achieves 98.78% of average detection accuracy which is the highest accuracy achieved only for corn disease detection from leaves with shorter training convergence times to the best of our consideration. Overall, this approach is advantageous and short time consuming and provides a dynamic way of detecting corn disease of the leaf. This will be beneficial for poor farmers from crop loss and lead our nation to support the digital agricultural system.