DAISY: Cattle-Mounted Biodiversity Sensors
Over the summer, 14 postgraduate students developed DAISY, an affordable system for monitoring farmland biodiversity, as part of their MRes in Sensor Technologies and Applications. This project tackles an important question in the agricultural industry: how can farmers most efficiently determine the quality and sustainability of their grazing pastures?
Farmers are motivated to maintain highly biodiverse fields because they provide more nutrients for livestock than monoculture pastures. However, actively managing the types of grasses, legumes and herbs across their farmland is a highly time-consuming process. At present, fields are sampled manually once every few years using quadrats, so results are often outdated and expensive to reacquire. Therefore, the agricultural research institution Rothamsted Research approached the Sensor CDT to automate their biodiversity monitoring techniques.
DAISY aims to address this issue with a custom (i) camera-based sensor, (ii) machine learning algorithm and (iii) user-friendly interface. Firstly, a low-power device captures geotagged photographs throughout the field. These cameras are mounted to the collars worn by cattle so that images will be taken in the field without human intervention. The device’s industrial design was heavily influenced by the day-to-day conditions encountered on farms. For example, its 3D-printed casing was made using durable sintered nylon to withstand physical shocks, and any ingress points were sealed using laser-cut rubber gaskets to prevent water damage in the rain.
Secondly, DAISY’s machine learning model identifies species such as grass, clover, bare soil and dung in each picture and evaluates how confident its predictions are. This model was trained and validated with images from online datasets and fields at Rothamsted Research in North Wyke, Devon. Thirdly, farmers can view the results via an online graphical user interface, where they can see the distribution of each species across their pastures.
In line with the UKRI’s open access policy, DAISY’s bill of materials, printed circuit board design and code have been published in an open-source Git repository, so other researchers can build upon this proof-of-concept system for real-world usage. In particular, this automated method can help more farmers access grants from government schemes like the Sustainable Farming Incentive.
During the project, each team member improved their technical skills, in fields such as electronics, machine learning and statistical analysis, and their soft skills, including time management, communication and problem-solving. Therefore, this project effectively prepared the students for any challenges they will likely face when completing their PhDs.
You can find out more about this year’s Team Challenge, including the benefits of using DAISY, how the system functions and how to navigate its graphical user interface, at daisysensing.com.
