Researchers from the University of South Australia and Middle Technical University in Baghdad have developed a smart irrigation system using an off-the-shelf digital camera and machine learning that identifies soil moisture content. A recent report from the United Nations states that many areas worldwide may not have enough freshwater to meet agriculture demands by 2050 if humans continue our current use patterns.
One solution to mitigate that problem would be to adopt more efficient irrigation practices, in which soil monitoring plays an essential role in terms of moisture. Incorporating sensors to guide irrigation systems would ensure watering is done at optimal times and rates. While sensors buried in soil can provide accurate soil/moisture readings, they’re also subjected to salts within the substrate and require specialized hardware and connections to withstand wide-ranging temperatures and corrosion. Thermal cameras are another solution; however, they can be expensive and become compromised by climate conditions such as fog, bright sunlight, and clouds.
The researcher’s smart irrigation system can overcome those hurdles by being cost-effective and able to monitor the soil in most conditions. The setup uses a standard RGB digital camera to accurately monitor soil conditions by “looking” at the differences in soil color to determine moisture content. The camera is connected to an artificial neural network (ANN) that uses specialized algorithms to recognize different soil moisture levels under varying atmospheric conditions.
“The system we trialed is simple, robust and affordable, making it promising technology to support precision agriculture,” states Dr. Al-Naji in a recently released paper on the system. “It is based on a standard video camera which analyses the differences in soil color to determine moisture content. We tested it at different distances, times and illumination levels, and the system was very accurate.” According to the paper, the system could be tailored to recognize soil conditions in nearly any location, allowing it to be customized for different ecosystems and climate conditions for maximum accuracy.