Computation Analysis of Multi-Class Plant Disease Diagnosis using Deep Learning Models

Horticulture is the art of cultivating fruits and vegetables, forming the base of both health and business. Despite replacing traditional methods, plant monitoring and disease diagnosis have even been the primary focus of modern-day farming as plant disease ignorance comes with monetary loss. Horticulture heading towards industrial Internet of Things (IIoT) 5.0 comes with smart methodologies to identify and treat plant diseases. With this objective, this paper emphasizes identifying a model which can well diagnose plant disease by diagnosing plant leaves with a high accuracy level. Three models namely ResNet50, DenseNet-161, and Inception V3 using Logistic Regression classifier have been compared based on the plant leaf image data set. Accuracy achieved for each model were Resnet-50 (96.9%), Densenet-161 (92.8%) and Inception Net V3 (90.2%) out of which Resnet-50 outperforms the other two.