KennelNet: A machine learning approach to postural estimation and behavioural monitoring in kennelled dogs. In partnership with Dogs Trust – The UK’s largest Dog Welfare Charity

This project provides an annual stipend of £19,237. 

Project advert

In the UK, 130,000 dogs are housed in animal rescue centres at any one time. Monitoring their health and behaviour is a major welfare challenge, as assessment is often conducted via labour-intensive manual observations by centre staff and veterinarians. Assessments may lack consistency, and behavioural testing can be stressful for animals. Automated video-based monitoring of dog activity levels and behaviour within kennels is a potential solution.

This project aims to revolutionise welfare assessment in kennels by developing an objective, easy-to-use video analysis tool, leveraging deep learning techniques. You will implement cutting-edge convolutional AI and deploy explainable techniques. Building on our ongoing work in controlled laboratory settings, where we have demonstrated the ability of deep learning to accurately track dog movements across various breeds, we will now translate this research into a practical tool deployable by non-specialists in rescue centres.

You will partner with Dogs Trust, the UK’s largest dog welfare charity, operating twenty-three rehoming centres across the UK and Ireland. Dogs Trust uses evidence-based approaches to improve dog welfare and actively participates in scientific research. Their extensive expertise, ethical commitment and nationwide influence will be invaluable in developing and implementing our AI-powered research to advanced animal welfare assessment.

Project aims and objectives

Aim:The proposed research project aims to develop an advanced, non-invasive machine vision tool for the assessment of canine behaviour and welfare in kennel environments.

Objectives:The research objectives are to:

1. Develop an automated video triaging pipeline using pre-trained object detection models to confirm dog presence and exclude humans in footage, optimising data storage and transfer.
2. Create “KennelNet,” a deep neural network for tracking dog posture using annotated video data, enabling detailed behavioural analysis across diverse breeds and environments.
3. Implement deep clustering techniques to identify discrete canine behaviours from posture data, enhancing explainability and validating the system against wearable sensor data.
4. Establish a real-time tracking and behavioural analysis system using lightweight networks and advanced filtering techniques.

Specific requirements of the candidate

This represents an opportunity to join the Faculty of Science and Engineering’s growing doctoral research community, committed to excellent research with impact. Successful applicants will be active researchers in our new state-of-the-art £117M labs and Dalton Building facilities, and will be supported to develop their skills as independent researchers.

The successful candidate will span the natural sciences and computer sciences, being supported by a diverse supervisory team incorporating animal behaviour (Dr Alyx Elder; Dr Alex Mielke), biomechanics (Dr Charlotte Brassey) and computer vision (Dr Connah Kendrick). The candidate will also work in close partnership with Dogs Trust (Dr Sara Owczarczak-Garstecka), the UK’s largest dog welfare charity. Manchester Met is a member of the Turing University Network, and the student will benefit from a range of training and development opportunities within the field of AI. Therefore, candidates should be self-motivated and driven to the creation of tech for good. Preference will be given to candidates with experience in computer vision, deep learning, and/or machine learning. However, there will be scope to tailor the focus of the project to match the specific skill set of an exceptional student, either from a biology or computer science background. We encourage candidates to apply even when they meet only a subset of the desired criteria.

Essential Criteria:

• Undergraduate degree in a related science field (including, but not limited to, biology, zoology, veterinary science or computer science and mathematics).
• Programming skills (in R or python).
• Experience with video data processing and analysis.
• Strong analytical and problem-solving abilities.
• Ability to work independently and as part of a multidisciplinary team.
• Excellent written and verbal communication skills.

Desirable Criteria:

• Post graduate degree in a related science field (including, but not limited to, biology, zoology, veterinary science or computer science and mathematics).
• Experience with machine learning and deep learning frameworks such as PyTorch or TensorFlow, demonstrated through a portfolio or GitHub account.
• Knowledge of toolkits such as DeepLabCut or SLEAP.
• Experience working with canines (professional work, volunteering, or personal commitment)
• Understanding of animal welfare and behaviour, particularly in companion animals.
• Demonstrated ability to manage large datasets.
• Experience with cloud computing and data storage solutions.

How to apply

Interested applicants should contact Dr Charlotte Brassey (c.brassey@mmu.ac.uk) or Dr Alyx Elder (a.elder@mmu.ac.uk) for an informal discussion. To apply, you must:

Complete the online application form for a full-time PhD in the Department of Natural Sciences (or download the PGR application form).

Complete the (PGR thesis proposal and a Narrative CV) form addressing the project’s aims and objectives, demonstrating how your skills relate to the area of research, and why you see this area as being of importance and interest. 

If applying online, you will need to upload your statement in the supporting documents section or email the application form and statement to pgradmissions@mmu.ac.uk.

Closing date: 14 October 2024. Expected start date: January 2025 for Home students and April 2025 for International students. 

Please note that Home fees are covered. Eligible International students will need to make up the difference in tuition fee funding. 

Please quote the reference: SciEng-2024-Dogs-Trust