Towards More Precise and Purified Store Traffic Data: Link People Counters

For the brick-and-mortar retailers, store footfall has been measured by deploying people counter devices for decades. The main categories of people counting systems include beam sensors, CCTV cameras and 3D/Stereo sensors. Each of them has their own pros and cons in terms of cost, accuracy, durability and scalability. The people counters are established as a commodity and people counting data integrated with sales figures offer the most basic benchmarking KPI for the retailers; conversion rate – share of visitors to your stores buying something. 

Our people counting solutions are supported with AI based algorithms. This allows us to train deep learning models to recognize common retail scene objects like shoppers, store employees, shopping carts, bags, products, shelves, etc. The trained AI models can be utilized for solving various troublesome technical limitations in people counting such as shadow detection, person detection and staff exclusion. For example, in Link Retail, we have already trained an AI model which can automatically detect and remove staff movements from the counting data. Our staff exclusion model is based on a holistic human behavior evaluation approach. This means that it does not require any tags, logo or sensors to be worn by the staff to identify them. The AI model analyzes the all-day activity at the end of the day and filters out staff activity from the data collected. The technology uses both computer vision and deep learning algorithms to detect and recognize humans. Our AI model is capable of discriminating human and non-human objects such as shadows, shopping carts, baby trolleys, etc. As a result, AI based people counting gives a huge accuracy boost for our retail footfall counting systems. Furthermore, it works 100% anonymously and GDPR compliant. Staff exclusion is just the first step for the AI algorithms evolving the whole retail analytics infrastructure into more precise and cost-effective systems. The best thing is that our sophisticated AI algorithms work with existing CCTV hardware without any hassle.

Store footfall is one of the most important KPI´s in retail today and it will be so also in the future. People counting systems powered by AI algorithms have just started to overcome the technology limitations for the footfall counter systems. In Link Retail, we have developed and deployed a few hundred systems utilizing our property AI models for our clients. Contact us today to get more information about our video analytics solutions.