Security Technology Executive

JUL-AUG 2018

Issue link:

Contents of this Issue


Page 32 of 83

www. • July/August 2018 • SECURIT Y TECHNOLOGY E XECUTIVE 33 Surveillance? AI and machine learning are key to supporting real-time responsiveness for video surveillance systems. Through machine learning , for instance, systems can be trained and adapted to identify nuances and differences in patterns, shapes, colors, sounds, vibrations and temperature, for example, which are all important and at times critical to help identify and detect issues in real time through applications like facial recognition that may be deployed for advanced identification, verification, search, prevention, response and rescue. Deep learning takes AI capabilities even further, analyzing video data to extract valuable insights. This is an intense process as you can imagine: Effec- tive deep learning involves higher computing power and can require thousands of hours of "training" with actual video footage to discern one human behavioral pattern. And this deep learning is tak- ing place all along the data path, from data in the surveillance video servers at the "edge" to server- based AI-enabled NVRs, to the cloud through Video Surveillance as a Service. The location of your data is also key, and the use cases will define this storage strategy. "Big data" is aggregated in the cloud to collect large volumes of data and analyze it over time, and can also be the data captured on NVRs in between the edge and the cloud. This is where deep learning functions operate, for example. On the other end of the spec- trum, "fast data" happens at the edge on cameras and other connected devices in the field that needs quicker, even real-time, analysis for emergency situ- ations. Take, for instance, a stadium hosting a large sporting or concert event: If surveillance detects fast movements (such as a car or motorcycle) approach- ing the venue in an unexpected and atypical man- ner, fast data allows the system to analyze the foot- age without any human interaction to flag urgent insights. Tampier: It used to be that recorded video was used primarily for forensics for law enforcement or » When it comes to security technologies, it's not a "one and done" approach because not only is the technology changing, but the threats are evolving as well. « — Rick Tampier, a Senior Dir. Sales & Product Strateg y at Red Hawk Fire & Security prosecution. What we're seeing today is that with innovations in the technology there are tools that can help us take action in real time to emergencies or other situations. One example is a large, multi-location company Red Hawk works with that recently began having parts stolen from the vehicles parked at their facili- ties. Digital cameras with analytics built-in can be used to identify someone crossing the perimeter of the facility combined with the ability to send an audio warning to 'talk-down' potentially averting a theft. If the situation escalates authorities can be automatically notified quickly. Many of our customers are interested in video analytics and remote video because these sophisti- cated cameras can serve multiple purposes, not just capturing clear images of the property, but being the eyes and ears at ATM locations or document- ing regulated employee practices like hand washing in healthcare facilities. There can be real value to improve business operations in addition to detect- ing , deterring and documenting security risks.

Articles in this issue

Links on this page

Archives of this issue

view archives of Security Technology Executive - JUL-AUG 2018