Another topic that has been gaining attention in video surveillance, natural-language video search, nevertheless ranked lower in our survey. This article takes a closer look at the survey results.
In asmag.com’s 2025 video surveillance technology survey, edge AI and multi-sensor cameras ranked high in terms of suitability and maturity. This comes as no surprise as both technologies offer various benefits for users and have become more technically mature over the years. Another topic that has been gaining attention in video surveillance, natural-language video search, nevertheless ranked lower in our survey. This article takes a closer look at the survey results.
Edge AI
Edge AI, where AI-based analytics from facial recognition to license plate recognition are embedded in the camera, received the highest scores in the survey. In terms of suitability and technical maturity, edge AI scored 4.42 and 4.03, respectively (1 being least suitable/mature and 5 being most).
"Edge AI has reached true maturity in the security industry, not just in concept, but in everyday use. We’ve been deploying edge AI for more than five years, and it’s now available across our entire camera portfolio, including our most affordable models. This means customers no longer need to choose between AI or no AI, intelligence is simply built in and ready to use,” said Philippe Henaine, Manager of Strategic Partners at i-PRO.
“What makes edge AI so ‘hot’ today is its practicality,” Henaine adds. “It delivers instant analytics directly on the device, reducing bandwidth and server costs, while keeping data processing fully local for enhanced privacy. Edge-based analytics can support both real-time alarm monitoring and forensic search, allowing operators to detect and review events with unmatched efficiency. In short, edge AI is a mature, proven technology that is reliable and scalable for every type of security project.”
Results of the 2025 video surveillance tech survey by asmag.com
Multi-sensor cameras
Also ranking high on the survey are multi-sensor cameras, which combine multiple sensors in one device. In terms of suitability, multi-sensor cameras scored 4.18, while for technical maturity they garnered 3.91. Again, this comes as no surprise as multi-sensor cameras can be quite beneficial: Instead of installing one camera for each direction, which can be costly, the user only needs one camera and one license to cover all directions. This makes multi-sensor cameras highly cost-effective for covering large areas such as warehouses and parking lots. As for technical maturity, the sensors in the camera now share a single systems-on-chip (SoC), lowering power consumption and simplifying network setup. Image stitching, which has long been a pain point due to visible seams or exposure mismatches, has now been significantly improved using advanced techniques such as real-time blending, HDR synchronization, and automatic exposure balancing. With these advances, we can expect demand and growth of multi-sensor cameras will continue.
VSaaS
VSaaS or video surveillance as a service ranks moderately on the survey. The technology received a score of 3.98 for suitability and 3.63 for technical maturity. Indeed, cloud-based video surveillance has its share of disadvantages and limitations. One of them is high bandwidth consumption, where high-resolution video consumes enormous uplink bandwidth – this makes VSaaS less ideal for remote area and multi-camera use cases. Further, real-time monitoring or analytics processing is difficult when video streams travel across the Internet from the end user site to the cloud. Finally, there is the issue of cost, as continuous recording from dozens of cameras can quickly become prohibitively expensive in the cloud. These factors make VSaaS less appealing compared to edge AI, which as mentioned has received much higher scores for suitability and maturity. Yet with the benefits of VSaaS, such as scalability, centralized management and reduced capital expenditure, we can expect the technology will remain popular and viable. And combining cloud and edge to form hybrid solutions will present an even better alternative for users.
Natural language footage search
In just a short time, ChatGPT, the chatbot assistant that can deliver almost anything requested by users via a natural language prompt and has pretty much rendered Google irrelevant, has become a global phenomenon. And this trend has spilled into security, where users can search for footage by way of natural language input, for example typing “Find a person wearing a red shirt and blue jeans between 2 and 3 pm,” instead of selecting different criteria one by one. The technology has gained strong attention this year and has been featured in various security trade shows. Yet on our survey, natural language footage search ranks moderately, scoring a lackluster 3.92 on suitability and 3.2 on technical maturity.
“Natural language footage search is an exciting innovation, but since many current solutions rely heavily on cloud-based processing, it has affected the adoption of the search technology. In the security industry, a large number of end users are still cautious about moving sensitive video data to the cloud for privacy, compliance, or infrastructure reasons,” Henaine said. “That’s why we’ve focused on delivering on-premises free-text search, that runs entirely without internet connectivity. This approach gives our customers and users peace of mind in their daily operations. Moreover, our technology ensures existing edge AI cameras can be updated to support these new capabilities, without requiring costly and troublesome hardware replacement.”
Vape detection
This year we added to our video surveillance survey vape detection which scored extremely poorly. For suitability it scored a mere 3.49, and for maturity, 3.09. This is understandable as video surveillance alone is limited in vape detection. For starters, detection accuracy is a key challenge. Video analytics can easily misclassify smoke, breath, and phones as vape or vape devices, resulting in false positives or negatives. Then there is the issue of privacy, as deploying AI video surveillance to detect vaping may raise privacy and civil rights concerns. A more ideal alternative would be to deploy vape/smoke detectors which can be installed in school bathrooms or other areas where vaping is banned yet privacy needs to be protected. Solutions offered by Verkada and Halo can be quite useful in this regard.