Implementation of Deep Learning for Image-Based Potato Leaf Disease Detection

Implementation of Deep Learning for Image-Based Potato Leaf Disease Detection

R. Ashish, M. Aditya, T. Samuel Raj, A. Vijaya Lakshmi

Abstract-
With the enhancement in agricultural technology and the use of artificial intelligence in diagnosing plant diseases, it becomes important to make pertinent research to sustainable agricultural development. Various diseases like early blight and late blight immensely influence the quality and quantity of the potatoes and manual interpretation of these leaf diseases is quite time-taking and cumbersome. As it requires tremendously a good level of expertise, efficient and automated detection of these diseases in the budding phase can assist in ameliorating the potato crop production. Previously, various models have been proposed to detect several plant diseases. In this paper, a model is presented that uses pre-trained models like VGG19 for fine- tuning (transfer learning) to extract the relevant features from the dataset. Then, with the help of multiple classifiers results were perceived among which logistic regression outperformed others by a substantial margin of classification accuracy obtaining 97.8% over the test dataset.

Keywords- feature extraction, logistic regression, inception V3, VGG16, VGG19, fine-tuning.