Page 24 - Security Today, March/April 2024

Getting Smarter About AI
Surveillance technology has change with new breakthroughs and enhanced remote monitoring
By Aaron Saks
The past few years have seen companies throughout the security and surveillance industry expand their use of AI, some rapidly adopting the technology and others dipping their toes in the water. Either way, AI has certainly moved beyond being an emerging technology to now being a proven reality, with demon- strated abilities to improve security camera imaging performance, enhance the accuracy of people and object detection, reduce false alarms and conserve recording and network bandwidth.
Surveillance technology has changed dramatically over the last few years with new breakthroughs in digital imaging and optics, enhanced remote monitoring capabilities, and more effi- cient methods of data storage and bandwidth management. The continued evolution of innovative solutions combining AI with on-board audio and video analytics is resulting in highly accurate object detection and classification, with fewer false alarms, plus the benefit of actionable data that can drive intelligent monitor- ing to enhance operational efficiency.
In 2024 and beyond, it does not take a tremendous amount of foresight to predict that AI will grow in use in every surveillance application across a range of markets. Also, AI is not a technol- ogy that operates in a vacuum; it affects several other types of surveillance technologies, including cloud-managed services and the use of data analytics.
With that in mind, the following are some of the more critical areas to keep in mind throughout the year and beyond:
Planum/stock.adobe.com
24
MARCH/APRIL 2024 | SECURITY TODAY
GENERATIVE AI VS. MACHINE LEARNING
AI has become a ubiquitous term for any robotic assistant or automation technology. But there are really two different areas. Machine learning is a subset of AI and refers to the ability of “machines” to learn over time. For example, think of when Netf- lix recommends other movies you might like, based on its analysis of your previous viewing habits.
Then there is Generative AI which has similar characteristics to machine learning, but also can “creatively” develop original content, ideas and images. In this case, think of Chat GPT or any of the open AI platforms coming to market.
In the surveillance world, generative AI can be used when col- lecting a dataset is too time-consuming or might include personal data. Generating a realistic yet graphical image with multiple variations like low light, snow or rain is much easier, and could become the basis for training AI models.
For example, in license plate recognition there are multiple variations like language, format and specialty license plates, and getting a real image of every single variation is almost impossible. Rather, if we were able to generate a graphical yet realistic license plate image with any variation. Now, we have a smarter AI model that can detect these variations without the need to collect datas- ets from each country and state.
This may become a larger trend as more companies rely on AI models to train their devices. For example, Hanwha Vision’s FLEX AI has the capability to convert any static object into data
ARTIFICIAL INTELLIGENCE