The Challenge
Lameness is recognised as the primary animal welfare issue in dairy farming. Nationally, 25% of dairy cows are lame at any one time. Although it has multiple causes, the key factor in most herds hoof health is disease such as digital dermatitis. Early detection and prompt intervention is critical to effectively control and treat lameness, which can cost farmers in excess of £300 per case. The UK Agri-Tech Centre, Hoofcount and the University of West England (UWE), funded by Innovate UK, launched the Hoofcount project in 2022.
Funded by Defra’s Farming Innovation Programme, this feasibility study brought together a consortium of partners to detect early signs of digital dermatitis lesions and lameness within dairy cattle.
UWE’s Dr Wenhao Zhang brings technical imaging expertise to the challenge. He said:
“Development of on-farm technology needs to be driven by fundamental research examining practical constraints in a bespoke way, in order to produce an innovative approach that is reliable, robust, and practicable. In this project, to solve the problem of object detection and classification ‘in the wild’, the opportunity to co-create this technology with different stakeholders and to inform design choices with the best farming practices and a wealth of inter-disciplinary knowledge is truly invaluable.”
The Innovation
The award-winning Hoofcount footbath was developed by a small Lancashire business and introduced to the UK market in 2012. Designed with simplicity in mind, it now has a sustained reputation in the UK as the market leader in effective and reliable footbathing.
Many hoof lesions are visible prior to lameness developing but can be difficult to see in practice and require specialist training to diagnose. The Hoofcount project is working on integrating machine vision technology and AI software for objective early detection. The aim is to realise algorithms, able to capture, filter, and analyse hoof images several times daily in a non-invasive way, detect hoof issues in the earliest stages and monitor for changes.
The Solution
Hoofcount identifies the animal via its RFID cattle tag, uses a powerful algorithm to image clean hoofs, and analyses the data to detect problems and recommend treatments. This non-biased information is sent to a portal or an app ready to provide farmers with a decision-making tool.
This project faced unique challenges in machine vision and learning due to the impracticality of manually annotating large datasets of hoof issues with variable appearances. Working farm environments add complexity with unstructured conditions, varying illumination, and unpredictable cow behaviour. Techniques were developed for robust, automated image data collection and processing to enhance machine learning model performance.
Reducing lameness in dairy cows has the potential to make a huge difference to the dairy sector in the UK and worldwide. Implementation of Hoofcount can:
- Reduce chemical use as a result of early detection
- Improve welfare standards
- Lead to better traceability
- Reduce carbon footprint
