Retail video analytics

People counting software

As a data science company, the BitRefine group helps retailers solve such tasks as precise pricing, targeted advertising, demand forecasting, customer churn prediction, recommender systems. All these tasks rely purely on available data. Here, video and corresponding recognition tools play an important role in collecting some of the required data. In addition to data collection, video analytics plays an important role in loss prevention as it allows detecting suspicious shoppers and alerting personnel in advance.

Traditional computer vision technology provided analytics tools that helped solve listed tasks to some degree. Today, thanks to artificial neural networks, retail video analytics became an incredibly powerful tool as now it lets you build your recognition system in such a way, that it will be recognizing any object you need.

Unfortunately, people haven’t yet invented a universal artificial neural network that will be able to recognize everything at once. Each artificial neural network is still trained to recognize just a certain number of objects. Some of the neural modules detect products on shelves, other people. To overcome this limitation BitRefine has built a multi-purpose platform where, depending on the current task, various neural modules get involved in the video processing pipeline. Some of the neural modules in BitRefine’s library are already pre-trained to detect standard objects, such as people, faces, bags, etc. In many cases though, retailers need some specific objects to be detected, classified or counted. BitRefine’s specialists prepare such customized neural modules by re-training them with sample images of target objects. After training completed, ready-to-use neural modules loaded into the BitRefine Heads platform’s processing pipeline and start extracting required data.

Retail video analytics software

Retail video analytics solves following standard tasks:

  • Visitors counter – based on human body recognition, this detector can effectively count people inside and outside the shopping area, where ceiling counters can’t be installed.
  • Shopper action recognition – based on gesture recognition that allows assessing shopper interaction with certain products
  • Loss prevention – based on face recognition. Faces of people that were previously recognized as thieves saved in a blacklist so that the system can notify personnel in advance.
  • Personalized advertisement – based on automated age and gender recognition.
  • Automated retail shelf analytics – based on automated detection of custom object on shelves.
  • Personnel tracking – based on face recognition.
People counting software reports view

BitRefine Heads platform converts video into structured data and saves it into internal DB. The reports section of recognition software provides seamless visual access to saved data. If the platform is used as a people counting software, it will show a chart with numbers of visitors on a timeline. If the recognition pipeline includes the face, age, and gender recognition it will show two different charts, separately for men and women. In addition, it will provide statistics on visitors’ ages.

Related Solutions

We’ve got more retail video analytics to share. Let’s be friends. But you first...