dashboard

RETAIL & FAST FOOD

Enterprise-level advanced video analytics system

We developed an advanced video analytics system that helps companies benefit from metadata, get expanded statistics visualized in a simple and structured way, and evaluate analytical results.

computer vision logo

Expertise

  • Data Science
  • Business Intelligence
  • Artificial Intelligence
  • Machine Learning
  • Computer Vision
blockchain logo

Technologies

  • Microsoft Power BI
  • Tableau
  • AWS Cloud
  • Amazon S3
  • Amazon Recognition
  • Amazon Relational Database Service
  • Amazon Redshift
  • OpenCV
  • OpenPose
  • Python
  • R

Terabytes of data could be leveraged far beyond mere security issues.
With cameras around, hypermarkets, cafes, and restaurants feel safe and know that in case of incidents, they restore a chain of events using video records.
Video control systems allow them to detect movements mainly or track the most basic events.

Meanwhile, most of retailers overlook the fact that massive amounts of accumulated data could be a ground for generating valuable insights with advanced video analytics solutions.
On top of that, they identify people’s behavior patterns and types of activities, but not only moves per se.

Challenge

Retailers are fully aware of the need to stay ahead of the competition in the industry. They need to track KPIs and accurately measure performance to achieve their sales and marketing goals.

The requirements are:

  • Calculate how much time clients spend on using self-service machines or self-checkout units
  • Provide statistics about the percentage of females and males in the total number of visitors
  • Identify the most popular age of the target audience and separate minors from adults
  • Track clean and dirty service areas/free and busy places
  • Control a drive-through service level thus detecting potential car accidents
  • Know the customer’s average stay time in-store or/and in a specific area
  • Get top visited store sections
  • Keep an eye on ‘idle’ visitors who attend stores regularly, but skip purchases
  • Explore correlations between the number of visitors and weather conditions
  • Define ‘principle buyers’ (either families as a unit or single attendees who have made a purchase)
  • Indicate a lack or absence of the most in-demand product on display to address the shortage in no time

All customers’ demands are related to the desire to improve business efficiency and optimize their clients’ purchasing experience.

Solution description

All the customer’s requirements turn into particular metrics.

The solution shows the clear picture for any period wanted.

  • How many visitors were there in September and what was the female percentage ratio?
  • Did a rush of customers depend on heavy snow or high pressure?
  • How long did it take to make an order via a
    self-service kiosk?
  • How many people a day leave a store without any purchase and who are they?

Management dashboard gives answers to business questions.
Our advanced video analytics solution gives the answers represented by handy reports and dashboards.

Major milestones

Our engagement started with uploading terabytes of video from cameras/surveillance systems to the cloud.
Then we got metadata after information processing, added them to a database, and visualized analytic.

Development аpproach

As our team dealt with enterprise-grade customers’ projects and huge amounts of data, we opted for a cloud solution. Our advanced video analytics system is deployed in the cloud, but we can make it on-premise as well at the customer’s request.
As for the ready-made tools our specialists selected for managing data, they all demand customization and fine-tuning adjustments to satisfy all customer’s requirements and provide quality results.

Specifically, we adapted OpenPose algorithms in order to reduce its HW requirements, optimize performance in real-time, and avoid invalid indicators.
We created a complex filteringsystem for advanced people tracking to separate the image of a person from real men or women.

Results

Our advanced video analytics solution based on machine-driven techniques lets customers understand their daily processes and operations at a deeper level.

Key takeaways from data processing results contribute to designing and measuring the most useful KPIs.

"Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry's standard dummy text ever since the 1500s, when an unknown printer took a galley of type and scrambled it to make a type specimen book"

Joran Lee (Head of Marketing)

ceo victor

Computer vision and data scientist with 15+ years of experience in R&D.
About 45 scientific publications and 15 patents. Worked for Samsung as a head of R&D department. Ph.D. in computer science. MBA degree from Open University. Extensive knowledge of image processing and pattern recognition.

CEO, Chief Data Scientist