An optimized Faster R-CNN model for Cassava Brown Streak Disease Classification

Abstract

The scientific community has shown considerable interest in plant disease detection and classification based on deep learning. In order to address these research gaps, this study proposes an optimized, fine-tuned model for the detection of Cassava Brown Streak Diseases. Casѕava is a vital Thai manufacturing harvest. Thailand is a pioneer in cassava production; therefore, a lot of cassava has been produced and exported. But, caѕsava infection could be the key to cut back caѕsava creation and immediately has an effect on growers' earnings. This research is to develop a model using an effective deep learning algorithm for cassava leaf disease detection. We split the classification into two phases, with Model1 and Model2. First model is used to do the cassava disease classification and second model for identifying the Cassava Brown Streak Virus Disease using VGGNet, AlexNet and Faster R-CNN algorithm. Furthermore, data augmentation techniques are employed during network training to improve the performance of the proposed network. The proposed model has been evaluated its performance using accuracy and confusion matrix. The experimental results demonstrates that our approach can accurately classify Cassava Brown Streak Diseases with an accuracy of 96% using Faster R-CNN.