An effective interest identification technique to enhance sales performance in supermarkets

Abstract

Surveillance Camera System is installed in the supermarkets mainly for security purposes. But the main idea of this paper is to use this surveillance camera system to improve the sales performance by targeting a particular stimulus (child) through marketing promotions. The owner of the supermarket monitors the entire store with the help of the security camera system. The owner suddenly finds an abnormal action in a stimulus (child) on looking at a particular product. On observation of the stimulus head and arm movements, the owner concludes the stimulus interest on that product which the parents refuse to buy. This scenario is implemented in this paper using live video analytics which identifies the abnormality. Action recognition is a technique that is used in the classification of actions present in the given video. The Bag of Visual Words Model is implemented for recognizing the action made by the stimulus. This model includes feature extraction, codebook generation and classification. The features from the stimulus such as arm and head are extracted using Speeded up Robust Features (SURF) algorithm. Codebook generation is done by K-means clustering and the histogram of discriminative features is generated and fed as input to SVM classifier which recognizes the action made by the stimulus (child) in order to identify the child’s interest factor on a particular product.