Deep Vision - A Machine Vision Framework for the Detection and Classification of non explored Pig Behaviour
School of Science, Engineering and Environment
- Funding
- Self-Funded PhD Students Only
- Application deadline
- Year-round applications
About the project
About the Project Traditional methods for monitoring pig welfare on large farms are labour-intensive and often fail to detect early signs of distress. This project aims to develop DeepVision Pork, an automated monitoring system that uses machine vision and deep learning to continuously analyse pig behaviour. The system will address the key challenge of accurately tracking individual pigs in crowded commercial pens, a problem that often hinders effective long-term monitoring. By processing video data in real-time, the framework will automatically classify a wide range of behaviours, with a specific focus on identifying early indicators of welfare issues such as aggression, tail biting, and lameness. The core objectives are to create a robust tracking algorithm, develop a highly accurate behaviour classification model, and produce a large, publicly available annotated video dataset to spur further research. The expected outcome is a proof-of-concept system that can provide farmers with timely alerts, enabling proactive interventions to improve animal health and welfare. This research represents a significant step towards the practical implementation of Precision Livestock Farming, promoting more sustainable and ethical pork production. Supervisors: Dr Ali Alameer Dr Taha Mansouri