Performance Analysis of Convolutional Neural Network for Plant Diseases Identification

Abstract

Our daily life starts with providing nutrition to our human body. A huge amount of food is provided by the agricultural sector. But always there isn’t 100% yield because of some issues like plant diseases, irregular rainfall, Natural disasters, etc. A major issue is plant diseases which are troublesome for this industry. An accurate and quick detection model is required for identifying the disease. In this paper, we have tested many classification algorithms for performance analysis such as EffecientNet-B0, GoogleNet, Resnext50 32x4d, and MobileNet-V2 on a GPU system. Various parameters have been taken into consideration for evaluating different classification models such as training time, training accuracy, and total loss to predict the best model which uses the least GPU cores and the result claims that Resnext50 32x4d gives higher accuracy.