Detecting Plant Diseases Using Image Recognition in Android Application

Abstract

The method suggested aims to identify plant disease and offers a solution that can be used as a disease defense mechanism. The Internet data collection is carefully separated and the various plant species classified and renamed to form a proper database, which also provides the test database consisting of various plant diseases used to verify the consistency and trust of the programme. Then we practice our classifier with training data and then we forecast the results with optimal precision. We use CNN (Convolution Neural Network) that consists of numerous predictive layers. The activation mechanism is the cornerstone of the CNN model since the non-linearity has an authentic classification artificial intelligence scheme. ReLu is one of the better functions for active input, but it has a benefit that the function derivative for negative values is zero and contributes to neuronal necrosis. New mathematical activation function to increase the device accuracy and efficiency on a TensorFlow framework is evaluated in comparison to acquired learning rate.