Detection of Illegal Dumping from CCTV at Recycling Centres

This paper describes initial work on a framework for automatic detection of illegal dumping from CCTV footage from recycle centres. Frames are seperated into foreground and background regions using a Bayesian approach that combines global motion estimates with image based information to generate a robust segmentation. The framework hence avoids explicit modelling and tracking of objects in the scene such as cars, people or rubbish bags. A feature extraction stage with diagnostics will be presented.