Product Placement Detection Based on Image Processing

Abstract

Product positioning in stores is of importance for the manufacturers. With this aim, the placement on shelves is done according to some documents named planograms. To check whether the placement is in compliance with the planogram, manufacturers need tools to gather information easily. In this work, using existing image processing techniques, we aim for the automation of planogram compliance control. We propose a novel technique for inventory management with the assumption that we can extract meaningful information from images using planogram context. The controller will be able to capture a photograph of the shelves and upload to the server to be processed by our system. The system is composed of three modules: 1) shelf detection, 2) product detection and 3) brand recognition. Basically, we make use of Hough Transform, Cascade Object Detection algorithm, and Support Vector Machines for the modules respectively. We test each module separately on a dataset we collected with cigarette products.