4-year PhD Studentship: Multi-camera machine vision of a whole cattle herd for assessing the impact of interventions for environmental sustainability

While dairy farming is receiving increasingly strong competition from plant-based alternatives, it remains an important agricultural option where grass is the most viable crop, such as Bristol’s civic region as well as a large slice of world landmass. A significant proportion of the global population, particularly in LMIC countries, will remain reliant on dairy production for economic and nutritional health. It is therefore essential to achieve sustainable dairy farming both resilient to climate change and to mitigate its high carbon footprint. With a £1M capital grant we are creating the ‘John Oldacre Centre’; the world’s most intensively monitored longitudinal cattle cohort to tackle the grand challenges in cattle health and sustainability [1] underpinned by our world-leading Animal Welfare & Behaviour research community to ensure that high welfare is built upon rather than sacrificed by sustainability [2]. A network of 25 CCTV cameras has been set up at Bristol Vet School’s Wyndurst farm with the goal of providing exquisitely-detailed tracking of individual animal activities, behaviours and interactions. In preparation, we have developed novel deep learning methods with the Department of Computer Science to individually identify cows [3,4], laying the groundwork to develop the automated video monitoring required for the proposed studentship.

Aims and objectives

Hypothesis: Optimising cattle breeding for reduced emissions and/or increased resilience to heat stress will not reduce health or welfare of the animals whilst having a significant positive impact on environmental sustainability.

Secondary hypothesis: The poorest performing animals contribute markedly to the environmental burden [5]. Automated video monitoring can realise the earlier detection of these individuals .

Technical objective: Creation of a multi-camera platform with machine vision classifiers for tracking the activities (loafing, lying, feeding) and social behaviours (face/body collisions, direction and nature of interaction) of the cattle herd for predicting individual-animal sustainability within a life-cycle assessment (LCA) model.

Methodology

The student will develop skills in AI and statistical modelling while learning livestock sustainability, animal science and welfare - a combination to underpin a future leading role in environmental sustainability in academia, industry or policy work. The core concept is a novel ‘virtual herd’ paradigm, where our herd will be split into sub-groups that can be tracked by the AI monitoring system while double-blind to farm staff and researchers. The sub-groups differ only by genetic selection strategy: (i) emissions-minimisation; (ii) heat-stress resilience; (iii) control group. After the underpinning AI is developed through extension of our deep metric learning work [3,4], the groups will be compared through implementation of an LCA model that predicts individual-animal sustainability given feeding time and energy expenditure (derived from the AI) and milk production/quality. Regressing AI parameters to health records will link early behavioural changes to potential welfare problems. We will then evaluate if the techniques translate to single-camera setups for translational impact across the dairy sector. Main supervisor and data scientist Dowsey will mentor Early-Career Researcher, secondary supervisor and dairy sustainability specialist Enriquez-Hidalgo. Held, Mullan and Burghardt will provide key expertise in fundamental behaviour, applied welfare science, and AI for animal biometrics respectively.

How to apply for this project

This project will be based in Bristol Veterinary School in the Faculty of Health Sciences at the University of Bristol.

Please visit the Faculty of Health Sciences website for details of how to apply