IoT-Based Crop Recommendation Using Deep Learning

Abstract

In India, the majority of people consider agriculture to be their primary occupation. The nation’s socioeconomic standing is mostly influenced by crop production and agriculture. The uneven use of fertilizers, defined as both excessive and insufficient amounts, which results in low crop yield and quality. The suggested system has been developed to provide soil analysis through sensor-based soil parameter measurement and observation. It combines the Internet of Things (IoT) with Deep Learning (DL). The system makes use of a number of sensors, including soil moisture and the DHT11, to monitor temperature, humidity, and soil moisture, respectively. After being collected by various sensors, the data is placed in a unique database and subjected to analysis using a sophisticated deep learning algorithm, specifically the Sequential model. These cutting-edge analytical methods aid in the creation of crop cultivation suggestions appropriate for the particular soil conditions noted. To test and confirm the model’s accuracy, activation functions such as sigmoid, Softmax, ReLU, and tanh were employed. The accuracy was further verified by contrasting our model with the Long Short-Term Memory (LSTM) model.