Plant diseases detection using an improved CNN model

The share of agriculture in GDP has been raised to 20 percent in the year 2020-2021. 70 percent of the rural household depends on agriculture. 82 percent of farmers are small and marginal growers. It's necessary to maintain agricultural growth. The prolonged plant diseases affect the growth of food products.There is much research work carried out to predict plant diseases. Deep learning methods are providing better performance on trained data but don't predict diseases during testing. This work proposes an improved CNN model to detect plant diseasesduring testing and to provide better accuracy. The proposed model is trained with different hyperparameters like learning rate,solvers, batch size, and number of epochs. The proposed work is compared with existing methods like NASNet, DenseNet, and ResNet-18. The Improved convolutional neural network model outperforms in terms of accuracy. It provides 98.10% of accuracy whereas NASNet provides 71.31% of accuracy, DenseNet provides 98.01% of accuracy and ResNet 18 provides 97.5% of accuracy