Advanced Video Analytics Adoption Is Lagging; How Can Camera OEMs Help?

At last year’s Consumer Electronics Show (CES), Bosch was recognized for its Gun Detection System, a video analytics-based solution designed to identify persons carrying a gun. According to Bosch, the Intelligent Video Analytics (IVA) product is built for the edge—the system does not require video data to be sent off the device, saving customers bandwidth costs and time to respond to an oncoming threat.

The system represents the video market’s shift to more complex, proactive video analytics workflows, characterized by edge recognition of specific objects beyond just people and vehicles. However, as many industry insiders have noted, though analytics capabilities have matured to include edge-based gun detection and other advanced use cases, there remains a gap in the market between what video analytics customers want and what they are willing to implement. Many video users still defer to basic motion and object detection workflows, even as they express interest in more complex, operational use cases.

Bosch’s and other camera Original Equipment Manufacturers’ (OEMs) push for advanced, onboard video analytics, when they were once so hesitant to get into the video analytics game at all, represents an industry-wide effort to bring a new generation of video analytics to the fore.

Though Bosch has a reputation for investing in video analytics, camera OEMs are not known for spearheading video analytics innovation. Until recently, camera OEMs did not view onboard, Artificial Intelligence (AI)-supported video analytics as a wise differentiator, given that its application varied widely across industries and training a model to fit all potential use cases was deemed too difficult.

Camera OEMs focused instead on improving their camera resolutions, hoping it would give them an edge over competitors. However, after Chinese camera manufacturers flooded the market with cheap cameras with sufficient resolutions, the devices quickly became commoditized, leaving OEMs like Bosch to pivot to intelligent on-camera video analytics to differentiate themselves and keep camera prices afloat.

In adopting these more advanced workflows, many OEMs are downstream of specialized video analytics software companies that have been developing in this area for a while. For example, ZeroEyes is a software company founded in 2018 that offers a gun detection video analytics solution as its main product. Scylla AI, another AI video analytics firm, also has a popular gun detection system. Konica Minolta, a technology provider closely partnered with camera manufacturer MOBOTIX, leveraged Scylla’s expertise in gun detection when adding a new visible weapon detection module to its platform in 2023.

Software and AI companies like these were the first to identify vertical-specific use cases for video analytics and are now likely shaping which workflows camera OEMs choose as they expand their analytics portfolio. As one industry insider told ABI Research, in their search for in-demand video analytics, OEMs will look for software companies that have already built an edge-based industry-specific video analytics solution, like Protective Personal Equipment (PPE) detection for manufacturing or liquid leak detection for oil & gas, and will either buy it or white label it.

This industry-wide effort from both OEMs and software companies to bring industry-specific video analytics to the market will likely improve the technology’s adoption, eventually persuading customers to mature past their basic analytics capabilities.

Currently, most video surveillance customers view advanced video analytics workflows as expensive add-ons without a clear Return on Investment (ROI). Camera OEMs are particularly well-positioned to change their minds, given that manufacturers own the hardware and can administer these complex analytics in ways that encourage their adoption.

Several industry insiders have noted that customers are more willing to use video analytics if the technology is automatically turned on and included in the camera’s regular functioning. This is likely how video users are persuaded to adopt basic analytics capabilities, like object and motion detection. Camera OEMs have already set these basic workflows on most cameras and allow customers to easily integrate these analytics with their back end Video Management Systems (VMSs). As OEMs follow the same pattern by placing more advanced video analytics on-camera with third-party integration support, advanced workflows like gun detection will become more normal and adoptable for the regular video user.

Camera OEMs, in pushing onboard analytics, will have to confront the need for verticalized solutions. Verticalization was one of the reasons camera OEMs initially avoided AI video analytics solutions—the use cases were too different by industry, and manufacturers were accustomed to tailoring their cameras to the entire market. However, customers are most likely to respond to video companies that understand their vertically-specific demands and can offer pre-built analytics solutions designed for these concerns.

New chip generations that support advanced AI models will help camera OEMs develop vertical-specific video analytics solutions. This will narrow down the effort and time it would take an OEM to create an industry-specific video analytics offering. OEMs are well-equipped to lead advanced video analytics adoption, given their resources and their established customer relationships. Incorporating more video analytics workflows, with the help of innovative software companies and advanced AI models, will allow camera OEMs to become even more influential in the video IoT world.