Transfer Learning approach to Sugarcane Foliar disease Classification with state-of-the-art Sugarcane Database

Abstract

In recent years, plant disease detection and classification systems have helped in better farming practices. With the advent of artificial intelligence, agriculture automation has seen innovative methods to mitigate risk and losses in farming. In this paper use of deep learning for sugarcane, disease classification is analyzed. Around 1470 images with 5 categories have thoroughly experimented. Transfer learning methods like VGG-16 net and ResNet are compared for an identical set of input parameters. The results obtained show with the limited set of datasets, transfer learning schemes can provide good results. VGG-16 Net and ResNet have shown accuracy around 83.00 % & 91.00 %, respectively.