A Deep Learning Data Augmentation Experiment to Classify Agricultural Soil Moisture to Conserve Plants

Abstract

Increasing water scarcity and frequent droughts pose a grave threat to agricultural production many areas. As a result, agrochemicals with potentially favourable effects on plants' ability to endure periods of low soil moisture are discovered. It’s classified the soil moisture using a Generative Adversarial Networks Architecture based on deep learning. On the next step, its implemented the Grey Wolf Optimization (GWO) technique for hyperparameter tuning of Generative Adversarial Networks (GANs) classifier. In this experimental, its mostly employed raw-based datasets to create high-quality cameras for train and test process. To detect and identify agriculture soil moisture image capture system was developed to record lively in land. Its described the Data Augmentation as a regularization based method for preventing overfitting. Additionally, this was utilised to duplicate the photographs by flipping, cropping, and rotating them. The classifier model can classify illnesses with relative ease. In the prediction of soil moisture at dry condition, an alert message is kept in a cloud-based Wireless Sensor Network (WSN) storage system, and then the GSM model is used to transmit the disease-affected message to the farmer's mobile device.