Transforming Agriculture using Artificial Intelligence Techniques - A Survey paper

Abstract

Use of modern technologies in the Agriculture sector is comparatively very less in India. Approximately 60% of people in India directly or indirectly depend on the agriculture sector but can contribute only 20% in GDP. Somehow India’s agricultural economy is undergoing structural changes. We have surveyed various problems faced by farmers and professionals working with the agriculture industry and listed problems faced in farming. We have also analyzed how technology can assist farmers for effective farming and getting healthy crop.  One of the major problems in the agriculture industry is disease identification and control. If it is not detected and controlled in the early stage, it can cause reduced yield and profit loss.  Other problems are soil health, weed control, weather condition and species recognition. Machine learning and deep learning technologies are used to solve these problems. Many of the researchers have used these techniques to solve the above problems in their research on different plants. Artificial Intelligence can also help farmers in yield prediction, fertilizer prediction or crop recommendation based on weather condition and soil health.  When Artificial Intelligence is used with IoT and Computer Vision, we can develop well-formed equipment or robots which can reduce the overhead of farmers. Farmers can monitor their crops from any of the places with an internet. In our survey, we have listed various Machine Learning algorithms that are used in agriculture. We have also compared ML models such as Artificial Neural Network, Support Vector Machine, K-means Clustering, Convolutional Neural Network and Bayesian Learning to get an idea which model will perform best in a given scenario. For any of the ML techniques, a precise dataset is required to train the model. We have listed a few publically available datasets used in various research.