Analyzing the Shopping Journey: Computing Shelf Browsing Visits in a Physical Retail Store
This study addresses the challenge of automatically recognizing in-store shelf browsing behavior by customers in physical retail environments. It proposes a machine vision approach leveraging overhead camera videos to extract 3D human trajectories and identify customer dwell events—termed “shelf visits”—in front of product displays. The method is calibrated using manually annotated data and validated across multiple store locations. Notably, this work presents the first model capable of effectively generalizing shelf-browsing recognition to unseen store environments. Furthermore, it establishes an analytical framework linking observed browsing patterns with actual purchase outcomes. Experimental results demonstrate that the proposed approach reliably detects cross-store browsing behaviors and reveals significant associations between specific browsing characteristics and subsequent purchasing decisions.