Using Computer Vision for Data Analytics for Retail Outlets

Abstract

Computer Vision has been widely used for analytics in several industries. The global retail market is $28 Trillion which is projected to grow more in the coming years. Analytics is very critical for this humongous market. Video analytics using Computer Vision now allows us to track micro details which provides valuable business insight. Stocking & Inventory, Consumer Behaviour and Safety are the key aspects of Retail Analytics. They construe the day-to-day functioning of the stores and provide us the statistics to analyze. This paper presents an approach to utilize Computer Vision to target the key domains and obtain useful data which can be applied for business intelligence. Examples of useful data obtained are person count, path tracking, spill and fall detection, etc. This processed data will yield business intelligence upon observation by the retailer. Machine Learning algorithms YOLOv3 and DeepSORT were utilized to create a robust retail analytics model.