In the future, the world is likely to face water and therefore food shortages due to reasons such as global warming, population growth, the melting of glaciers, the destruction of agricultural lands over time or their use for different purposes, and environmental pollution. Although technological developments are important for people to live a more comfortable and safer life, it is also possible to reduce and even repair the damage to nature and protect nature itself thanks to new technologies. There is a requirement to detect abnormal water usage in agriculture to avert water scarcity, and an electronic system can help achieve this objective. In this research, an experimental study was carried out to detect water leaks in the field in order to prevent water losses that can occur in agriculture, where water consumption is the highest. Therefore, in this study, low-cost embedded electronic hardware was developed to detect over-watering by means of normal and thermal camera sensors and to collect the required data, which can be installed on a mobile agricultural robot. For image processing and the diagnosis of abnormal conditions, the collected data were transferred to a personal computer server. Then, software was developed for both the low-cost embedded system and the personal computer to provide a faster detection and decision-making process. The physical and software system developed in this study was designed to provide a water leak detection process that has a minimum response time. For this purpose, mathematical and image processing algorithms were applied to obtain efficient water detection for the conversion of the thermal sensor data into an image, the image size enhancement using interpolation, the combination of normal and thermal images, and the calculation of the image area where water leakage occurs. The field experiments for this developed system were performed manually to observe the good functioning of the system.