Detect-o-thon: Identification of infected plants by using deep learning

Abstract

In this paper, model detect-o-thon uses deep learning and image processing to detect the plant diseases using various rigorous methods as mentioned as deep learning, where in convolutional neural networks is used and this plays a vital role in visual imagery. In this paper 56,725 images have been taken in total for training and testing the model among which 42312 are training images and 14413. TensorFlow is used to extract some of libraries as well as for image processing. Keras has dominant role in the whole paper which can be used for prediction of feature extraction and fine tuning hyperparameters, constantly interacting with the deep learning at the core of the process to detect the infected plants to get the best out of it. Mainly CNN is used for computing and image processing where different layers create using batch processing have helped the proposed model in achieving 96.84% of accuracy.