Case Studies

Omnigo

Search for scammers and other unwanted individuals by blacklisting using the face recognition module in the Las Vegas casinos

About Omnigo

Omnigo (Canada) offers the world's leading software solutions for public safety, incident reporting and security management.

"Unwanted Visitor" is one of the solutions that provides face recognition from the casino's blacklist and is installed in 350 casinos, including most casinos in Las Vegas.

Website: www.omnigo.com

Problem Statement

The company was looking for opportunities to improve recognition accuracy rates and reduce false positives in order to increase the efficiency of the "Unwanted Visitor" service. At the time of contacting 3DiVi, the Company used the other vendor's face recognition algorithm.

Solution

In early 2018, Omnigo Vice President began testing alternative face recognition algorithms, including 3DiVi.

March 8, 2018

Omnigo tested the 3DiVi algorithm on a database of 4,000 people and 700,000 test images. The preliminary results were encouraging and the client decided to move to the pilot stage on real data:
I am 60% through our testing of your SDK and it is looking really good. I have 4000 enrolled images and 700,000 test images from one of our Casino customers, which are their entries from a few weeks.
The preliminary results look like you will come out with much better results.
I am excited about this and would like to start to talk more about our working relationship.
Soren Frederiksen, Vice President - Innovation Lab

July 5, 2018

Omnigo Vice President has completed testing the 3DiVi algorithm with one of Omnigo largest customers:
I have concluded 3DiVi Face SDK testing with one of our biggest customers and I am happy to say that the results are really good. We tested 1.05 million images of people entering two casino's over a two week period. We had the old system results which are currently running as a production system to compare them to.

3DiVi Results (algorithm v 7.3)
  • 21 people recognized from a hotlist of 21,476 people
  • 1,831 alerts were rejected
Other vendor results
  • 25 people recognized from a hotlist of 21,476 people
  • 6,255 alerts were rejected

This means that in the 2 week test period with real patrons on a hotlist we found approx. 8 times as many people with 30% of the false alarms."
Soren Frederiksen, Vice President - Innovation Lab

November, 2018

We have completed the first phase of the live pilot with our customer and we are getting ready to deploy 3DiVi live in the end of January next year.

We got really good results in the live pilot testing compared to their current solution:
  • People recognized: Improved by 4.3 times
  • False alarms: Reduced by 70%

What we are doing between now and the end of January is hardening our software for production and testing the scalability of the solution. We are currently installed on two pilot servers, but will be moving on to production servers in early January.
Soren Frederiksen, Vice President - Innovation Lab

January, 2019

3DiVi Face Recognition SDK is deployed in a live system.