Automated weather forecasting and field monitoring using GRU-CNN model along with IoT to support precision agriculture

Abstract

India's agricultural sector is dwindling, which has an impact on ecosystem output. Smart farming increases crop productivity by reducing waste and making effective use of fertiliser. Various Machine Learning techniques together with IoT enabled are developed for precision agriculture. However, the existing techniques face difficulties in forecasting weather and predicting disease accurately. In this work, a system is developed to automate weather forecasting and field monitoring using sensors. Wireless transmission is used to transfer sensor data to a web server database. In the first step, the collected data is pre-processed using normalization and mean based missing value imputation to convert the raw data into meaningful data. Pre-processed data is then fed into Gated Recurrent Unit (GRU) for forecasting the weather condition. In the second step, crop images acquired from sensors for disease prediction are pre-processed using Adaptive Gaussian Filter for noise removal and Dynamic Histogram Equalization for contrast enhancement of image. Pre-processed images are then fed into ResNet50 for feature extraction and classification. Using these predicted data, in case of any horrible weather conditions and soil conditions, remedial action will be automatically taken by the systems. On the other hand, regarding the information about horrible conditions and in presence of pests and diseases, an alert message will be sent to the farmers. The proposed method is tested with several metrics which attain better performance like 94% accuracy for weather forecasting and 98% accuracy for field monitoring. Thus the proposed IoT and deep learning based model can support farmers to achieve a high quantity of crop production in lesser time.