Plant Disease Identification and Classification Using Convolutional Neural Network and SVM

The economy of agricultural countries depends mainly upon agricultural production. Crop production is a medium of livelihood for most of the population. Crops can be affected by diseases owing to various factors such as climate change, pests, etc. that can damage the crops heavily. Multiple systems have been proposed to detect diseases in plants at an early stage. The existing plant disease identification methods are tedious and error-prone as they require handcrafted feature extraction and segmentation. Due to which, the Convolutional Neural Network (CNN) based automated and robust methods are being employed in different research areas. This manuscript explores plant disease identification and classification in leaf images via Deep Learning (DL) based and Machine Learning (ML) based algorithms. The leaf images are initially resized, segmented and then supplied to CNN models such as AlexNet and VGG19 to extract deep features. These features are then classified via an ECOC based SVM classifier. The system achieved the highest accuracy of 98.8% via AlexNet and 98.9% via VGG19. Our proposed method outperformed existing plant disease classification approaches and can be used by farmers to detect diseases in various crops.