Developing Algorithms for a Multi-camera System to Locate People Within a Store
This internship project aims at creating a system prototype using several surveillance cameras in order to measure customer traffic inside stores. At the moment, most counting methods use sensors that do not differentiate between repeat visits by the same individual and single visits by different individuals. Moreover, these methods cannot always distinguish between potential customers and store employees or security personnel, thus reducing the quality of the counting measurements. The use of cameras should rectify these problems while providing with increased precision the localization of customers within the point of sale so that their shopping behaviour can be analyzed by indexing all the areas they visited and the time they spent in each store section or aisle. With the use of multiple cameras, the precise position of customers can be obtained while insuring all sections of the store can be viewed even in the presence of various obstructions such as store furniture or other objects. This research will draw on mathematical methods from the fields of computer vision and video surveillance.