Case Study: University of Glasgow studies litter patterns with Roboflow computer vision

Studying Links Between Litter and Socio-Economic Factors with Computer Vision

A postgraduate student at the University of Glasgow was exploring correlations between litter in urban areas and socio-economic factors. The challenge was to gather enough data on littering at a city-wide scale without the need for extensive and time-consuming fieldwork to enable this statistical analysis.

The student used Roboflow to prepare and augment his custom image dataset, which was then used to train a YOLOv5 computer vision model to detect litter. Using Roboflow allowed the model to identify 7,732 pieces of litter across 37,300 city images. While the study concluded no significant statistical relationship between litter and deprivation factors, it proved the feasibility of using computer vision for large-scale litter analysis and provided a model that could be leveraged by local governments for further research and policymaking.

University of Glasgow

Gary Blackwood

University of Glasgow