
OinkScope – A Computer Vision-based Spatial Intelligence tool for Monitoring Behaviour of Group-Housed Pigs
Project overview
Participants
More related research
Short summary
Video recording of ethological experiments creates a large bulk of footage to be manually annotated in order to extract the data.
Video recording of ethological experiments creates a large bulk of footage to be manually annotated in order to extract the data.
This often creates a critical bottleneck in data processing and can limit the time available for analysis and delay reporting of the results. To meet the growing demand for accurate automated behaviour monitoring, we present a new tool for pig behaviour monitoring.
OinkScope is a novel computer vision (CV) algorithm that integrates real-time object detection, customizable region-of-interest (ROI) monitoring, and automated heatmap analysis. At its core, OinkScope relies on Detectron2 object detection and segmentation framework built on the PyTorch ecosystem.
By uniting customizable ROIs, scalable deep learning models, and an intuitive GUI, OinkScope provides insights into group dynamics, crowding patterns, and resource utilization. Ultimately, it offers a powerful yet accessible toolkit for automatic and real-time behaviour analysis — showcasing the transformative potential of modern CV systems in ethological research.