Robust visual analysis for planogram compliance problem

Abstract

This paper presents a novel visual analysis based framework for automated planogram compliance check in retail stores. Our framework provides an efficient and convenient solution for ensuring planogram compliance by real-time analysis of the shelf image acquired in freehand manner. We present a novel application of Hausdorff metric for occupancy computation in product shelf images. Subsequently, we present a robust solution for product counting which applies robust row detection algorithm, and exploits texture and color feature for accurate counting. In this context, our system addresses the most general scenario of multiple varieties in single product type. The empirical validation of our framework is demonstrated on range of real-life images from stores located across different geographies, where it has achieved satisfactory and encouraging results.