Developing hybrid AI on wearables and video for fine-grained cattle behavioural monitoring

About the Project

Background:

Do you want to help revolutionise how we understand animal behaviour and welfare? Social network data is now a powerful tool for studying both human and animal behaviour, and this project offers the chance to contribute to cutting-edge research in this field. At the University of Bristol, we have built the John Oldacre Centre, where advanced technology like 50+ video cameras and artificial intelligence (AI) are used to monitor and track cows in a dairy farm setting [1]. This has allowed us to study subtle changes in social behaviour, which could help us predict diseases and improve animal welfare.

Aims and objectives:

Monitoring cows in open fields presents a new challenge - video footage is not practical when they roam outdoors. Based on our experience predicting declining health with animal-worn accelerometers [2], we have now developed our own bespoke and fully programmable wearable sensor and are excited to use it for new research. In this exciting project, you will help develop a novel system that learns to combine video monitoring with wearable sensors (such as collar and leg devices) to better track cow behaviour in both indoor and outdoor settings.

Methods:

You will spearhead the creation of hybrid system that will use AI to automatically link video data with wearable sensors through deep learning AI-based transformer models, improving our ability to understand animal behaviour wherever they are. Concepts you will work on particularly include transfer learning of behavioral classifiers from video to accelerometry data. You will also explore how indoor and outdoor environments affect animal behaviour and how to translate the system into practice, including how to optimise camera and sensor placement to gather the best data.

Key references:

[1] https://doi.org/10.1016/j.compag.2021.106133

[2] https://www.biorxiv.org/content/10.1101/2020.08.03.234203v4

Supervisors: This project is a collaboration between data science, machine vision, and animal science and behavioural biology experts at the University of Bristol (Prof Andrew Dowsey, Dr Laszlo Talas, Dr Daniel Enriquez-Hidalgo and Dr Suzanne Held) and Rothamsted Research’s North Wyke Farm Platform (Prof Paul Harris), a large-scale grazing research platform investigating the future of sustainable livestock farming.

Whether you have a background in computing or the mathematical sciences and are interested in machine vision and deep learning, or come from biosciences and want to develop new skills in AI, this project offers a tailored training package to support your development. Join us in exploring how cutting-edge technology can transform our understanding of animal behaviour and welfare!

Start date: September 2026

How to apply:

Before applying, please carefully read the information on the prospectus Veterinary Sciences | Study at Bristol | University of Bristol and make sure you meet the eligibility criteria and have all the documents listed in the PhD and MSc by Research Veterinary Sciences Admissions Statements. A research statement and supervisor support form are not required – please add a blank sheet of paper when these documents are requested.

To apply for this project go to http://www.bris.ac.uk/pg-howtoapply and on the “Start your application” page select “Veterinary Science (PhD) (4yr)”. Select Sept 2026 start.

Contacts: please contact fohs-pgadmissions@bristol.ac.ukwith any queries about your application.

Please contact the project supervisor for project-related queries: andrew.dowsey@bristol.ac.uk

Funding Notes

This project is open to applications from (home and international) students who can self-fund or who have a scholarship.