Page 32 - Security Today, March/April 2023

The Core Value Proposition By Shawn Kermani
AI assisted analytics enhance customer site awareness, both from a security point of view, but also from an operations point of view. AI-based cameras can also identify unique attributes of the objects it detects. This allows analytics to be further defined and customized to the end users’ exact needs.
If someone is loitering near the back door after hours, an AI-based camera can send a notification directly or via VMS to operators. If a queue at a register gets too long, it can notify the store manager. AI- based cameras can accurately count people entering a store so sales conversion rates can be calculated. It can also make evidence more useable and searchable. Thousands of petabytes record per day worldwide, which is unusable since it is too much for anyone to go through. With the additional descriptive metadata captured by an AI camera, a VMS can find all occurrences of a person in a red shirt and white shoes carrying a bag in just minutes, if not seconds. AI enhances securi- ty and allows operators to do more with less. It also provides business intelligence features that can have a positive impact revenue.
EVALUATING AI SOFTWARE
While AI-based solutions are everywhere, the implementation of the technology can vary widely between manufacturers. The powerful descriptive metadata captured from an AI-based camera is only useful if it can be interpreted and displayed within a VMS or NVR.
It is imperative to ensure that the VMS supports all the data from any camera sys-
Machine and deep learning algorithms are everywhere in our lives. Masquerad- ing as AI, they are only in their infancy. Have a con- versation with a ChatGPT chatbot, and it becomes clear just how far we have come in a short time and how far we have to go. For security professionals, it is no longer if, but when will AI-based analytics be standard on every surveillance camera in the market.
Predictive analytics and AI are en- abling security professionals to adopt a more proactive stance versus a traditional reactive one. AI-based object detection in network cameras, coupled with in-built analytics, has the capability of preventing or interacting with events in real time, not just documenting evidence of something that has occurred in the past.
The same technology behind self-driv- ing cars deployed at the edge in i-PRO cameras to provide operators with real- time situational awareness. AI is driven by numerous use cases, which is one of the reasons it is evolving so quickly. If a vehicle is loitering outside a jewelry store at 3am, an AI system can help predict that an undesirable event could soon take place and warn operators to take a closer look.
AI and analytics change the core value proposition by going beyond security and providing valuable insights into operations and sales for customers as well. Cameras have become IoT smart sensors that can mea- sure the flow of people and vehicles around a facility. Loitering, while undesirable from a security standpoint, becomes a positive indicator for marketing and sales profession- als when potential customers are pausing to look at a new product on display. Having the same camera that looks for shoplifters also can notify staff when stock replenishment is required represents tremendous opportunity to enhance operations and efficiency.
TURNING CAMERAS INTO SENSORS
AI-based object detection in network cam- eras make analytics more accurate and error free. Prior generations of analytics reacted to pixel-based motion that would include triggering when passing shadows or wind-blown objects moved in a scene.
The KonG/Shutterstock.com
32
MARCH/APRIL 2023 | SECURITY TODAY
tem you choose to deploy. For example, i- PRO’s Active Guard plug-in works seam- lessly with both Genetec and Milestone VMSs. Algorithms based on machine and deep learning are very useful for recogniz- ing vehicles and humans and detecting attributes such as vehicle/clothing colors, gender, age, objects being carried, glasses, hats, bags and more. Having this addition- al information makes motion analytics like line crossing, direction and loitering almost entirely error free.
Color is a critical component of suc- cessful AI analytics. It is also important to note how many attributes can be described for objects recognized since this directly affects the relative success of any search. For example, some cameras can detect shoe color, while others cannot.
To keep AI analytics functionality work- ing at night, colors that define objects need to be maintained. The goal should be to keep a camera capturing color mode as long as possible. This requires the best low light performance available from a lens/sensor combination. Make note of the low light ca- pabilities of any camera considered for AI- based use cases. Once a camera turns on IR to illuminate a scene, color will be lost since most cameras automatically change to black and white mode that effectively reduces the effectiveness of any AI-assisted analytics.
AI technology can to inform the cam- era’s image processing system about which objects in an image, such as a face, are im- portant to capture with the highest clar- ity. Look for features such as advanced
ARTIFICIAL INTELLIGENCE