Security Hardware and Technology

How to Add AI-Powered Video Analytics to Any Camera and Optimize Security & Safety

The current methods of actively protecting a company, city, or university’s physical spaces, people, and assets are costly and largely ineffective. Thankfully, Artificial intelligence (AI) and its application in video analytics has transformed the way live-monitoring and post-event investigations are done — with results that not only save time and resources, but lives. Additionally, whereas legacy video analytics solutions typically only work on video from fixed cameras and in stable lighting environments, new, cutting-edge products like Vintra’s real-time analytics solution, FulcrumAI, works on all types of footage, from both fixed and moving cameras, and with variable scene and lighting conditions. So, why video analytics, and why now?

oonal / iStock / Getty Images Plus / Getty Images

Outdated Security Practices

Guns, guards, and gates, the holy grail of traditional physical security, are no longer adequate tools. Relying on a call to the police by a security guard or an employee in the heat of the moment during, say, an active shooter situation is hopeful, at best. Through AI-powered face recognition, and the creation of safelists & blocklists, the shooter’s detected face, either as an unknown person or a known threat, can instead trigger an alert to security staff the moment the face is detected on any camera. Why invest in video analytics when you already pay people to monitor live camera feeds and review video?

Click here to see FulcrumAI in action

Huma Limitations are, well, Limiting

The stats on the effectiveness of humans’ ability to monitor and review video are sobering, at best. The typical central monitoring station is staffed by a handful of security operators watching anywhere from 2 – 5 screens. Humans simply are not built for this kind of sustained focus the job requires. After only 22 minutes of sustained video review, humans lose up to 95% of visual acuity and experience “video blindness”. One researcher did a study on how accurately security operators could detect a person carrying an umbrella on a busy street when viewing differing amounts of monitors. The study found that observers viewing one, four, six and nine monitors had accuracy detection scores of 85%, 74%, 58% and 53% respectively in identifying the only person in the crowd holding an open umbrella. Imagine if that umbrella was a weapon instead?

The Solution?

Through an AI-powered solution, like FulcrumAI, security operators and investigators can easily search video and set real-time alarms around face detections and recognition, weapon detection, person detection, gender, age, color of clothing, direction of travel, vehicle, vehicle subtype and color, object detection, and major gesture movements.

Click here to download Vintra’s Enterprise Security White Paper.

Thanks to massive advances in computing power and cutting-edge AI that blurs the line between fiction and reality, it is possible to implement systemic changes in a security or investigative team’s operations and focus, reduce wage-spend on security guards, repurpose man-hours for higher-level work, and transform real-time monitoring into intelligent monitoring with one cost-effective solution running through your existing camera network.

In addition to real-time watchlists powered by face recognition, a few other capabilities that improve the security and safety of physical spaces and facilities of all types include:

  1. Intrusion alerts: Set alarms and alerts around entry and egress points, rights-based access, and more.
  2. Boundary crossing: Powerful boundary crossing alerts to supercharge response times.
  3. Loitering: Parameters can be set around “time spent loitering” and location. Is someone on a smoke break, or are they about to break in?

What’s Next for Video Analytics?

The next horizon in video analytics is users creating their own case-specific algorithms. Whether it be a logo, a uniform, specific objects, a unique piece of equipment or hardware, users will soon be able to easily create their own detections in Vintra’s industry-first Analytics Foundry. Vintra’s solutions have been built on top of our own deep learning-powered algorithms, purpose-built for the safety and security industries, which enables users to not only search and monitor any type of video accurately, but to truly sit in the driver’s seat and optimize the solution to best fit their needs.

Conclusion

The best thing about AI-powered solutions is that they thrive at doing the boring and repetitive, but important, things that we as humans struggle to do efficiently or effectively. Remember when we used to stand on the corner in the rain and wave down speeding taxis? Now cars magically appear at our front door thanks to AI, ready to take us wherever we need to go. The same can be said for live monitoring of security video feeds and video review. The risks associated with protecting your people, spaces, and assets are too high not to augment the process with AI-powered video analytics. At Vintra, we strive to provide flexible, intuitive, force-multiplying tools that our customers can easily integrate into their existing security and safety processes.

1 http://www.cs.nott.ac.uk/~pszcmg/G64IDS/isd-dissertations-08/nxd07m.pdf
2 http://www.securitysa.com/article.aspx?pklarticleid=3313
3 https://blog.openai.com/ai-and-compute/

 

Print