Image-Based Plant Disease Detection by Comparing Deep Learning and Machine Learning Algorithms

Abstract

Plant diseases area unit the most issue two-faced in agriculture. As population can increase, the assembly of plants in addition can increase and due to plant diseases it's going to have a control on the assembly of food.The traditional methodology used for illness detection is knowledgeable visual observation. but it's very sophisticated to go look out the illness manually as a result of the time interval and knowledge of the plant's diseases. So, it had been necessary to develop a system that detected the illness in less time and value effective manner.We discuss the employment of machine learning and deep learning to sight diseases in plants automatically.Using a public dataset of fifty four,306 photos of pathological and healthy plant leaves collected below controlled conditions, we have a tendency to tend to coach a deep convolutional neural network to identify fourteen crop species and twenty six diseases (or absence thereof). The trained model achieves academic degree accuracy of 9ty nine.35% on a held-out take a glance at set, demonstrating the practicability of this approach. Overall, the approach of coaching job deep learning models on additional and additional large and publicly out there image datasets presents a clear path toward smartphone-assisted illness identification on a huge world scale.