Automatic Irrigation System Based on Computer Vision and an Artificial Intelligence Technique Using Raspberry Pi

Abstract

Efficient irrigation water use directly affects crop productivity as demand increases for various agricultural products due to population growth worldwide. While technologies are being developed in various fields, it has become desirable to develop automatic irrigation systems to reduce the waste of water caused by traditional irrigation processes. This paper presents a novel approach to an automated irrigation system based on a non-contact computer vision system to enhance the irrigation process and reduce the need for human intervention. The proposed system is based on a stand-alone Raspberry Pi camera imaging system mounted at an agricultural research facility which monitors changes in soil color by capturing images sequentially and processing captured images with no involvement from the facility’s staff. Two types of soil samples (sand soil and peat moss soil) were utilized in this study under three different scenarios, including dusty, sunny, and cloudy conditions of wet soil and dry soil, to take control of irrigation decisions. A relay, pump, and power bank were used to achieve the stability of the power source and supply it with regular power to avoid the interruption of electricity.

Keywords

• automatic irrigation
• computer vision
• random forest algorithm
• Raspberry Pi
• color image analysis