AI Case Study

Object and anomaly recognition within videos

Project description: Intelligent video monitoring system that knows what should be seen by the camera and raises an alarm when something seems off. The system can alert the management if a particular set of objects is detected on the video

Outcome: increased safety of controlled environments and decreased the response time drastically.

Background

There are few things as important as public safety, and we are proud to contribute to it.

Our client provides security services for public events where lots of people show up at the same place. There are certain objects that should not be seen in this type of events, and their presence must be responded to quickly. Typically a large staff of security officers has to monitor the space through cameras. Our client wanted to increase the reliability of the process by adding AI eyes in addition to human eyes.

Approaches we used

Analysis of individual frames from the video feeds
Analysis of changes from frame to frame

This way we could identify both known threats found on still images and unknown threats based on dynamics of data. If the profile of changes on the screen did not match the safe patterns, an alert was issued to a security officer to review and address.

Challenges we encountered

There are many factors that affect the AI's image parsing capabilities.
Camera model, resolution, number of frames per second, and lighting significanly influenced the quality of analysis and reliability.

We had to test a huge variety of scenarios to ensure that our solution covers majority of real life situations.

This was one of the projects that seemed somewhat simple initially, but turned out to be much more complex than any of the stakeholders anticipated.

Wrapping Up

Our client managed to increase the reliability of public safety monitoring, and the average incident response time decreased by 50%.

Unlike humans, AI does not get tired or lose attention. We managed to take advantage of this and make public spaces safer.

Team

1

Project Manager
3

AI Specialists
2

Back End Developers
1

Front End Developer
3

Quality Assurance Engineers
1

DevOps

7840

hours spent on project

Let's get started

To coordinate next steps please contact:

Mail: contact@zfort.com

Tel: +1 202 9602900

ZFort Group - Your reliable partner