Intel Acquires Video Analytic CompanyBy John Honovich, Published on Nov 22, 2010
In a potentially good sign for video analytics, but not for its security application, Intel has confirmed acquiring [link no longer available] a video analytic startup named CognoVision.In this note, we contrast using video analytics for digital signage vs. security / surveillance.
The digital signage market is fairly large (about $4 Billion and growing rapidly) though still smaller than surveillance. A big problem for signage is ROI justification and optimization - it is hard to determine the relative success (or failure) of displays.
Video analytics are viewed as a potential solution to measure and optimize digital sign displays. For the concept, see Intel's whitepaper on digital signage [link no longer available]. A few quotes capture their approach:
- "Gauging effectiveness traditionally entails poring over sales figures, interviewing consumers and oldfashioned observation, all of which can be tedious, time consuming and, ultimately, inaccurate."
- "Video analytics distinguishes how much attention people put toward digital signage. As someone approaches a monitor, cameras focus in, capturing a digitized impression of the signage viewer."
- "The technology is able to discern a person’s gender, race, approximate age and, based on the contours of the person’s face and positioning, just how long he actually looks at the screen"
- "Feedback produced by analytical impressions empowers businesses to make a host of changes on the fly. For instance, they can spot immediately what messages or ads are effective with customers"
The basic premise that tracking is hard and analytics can help measure make sense. The specifics are certainly more challenging - especially when it comes to camera angle and positioning and the accuracy of performing more sophisticated tracking such as gender, race and age.
Two positives that this application has over video surveillance are:
- Trend analysis rather than incident response: The system can be mistaken on any particular instance as long as the general trend it measures provides meaningful information. In security, this is much less tolerated as a missed incident can be catastrophic plus each data point (i.e., alarm) needs to be reviewed in real time.
- Revenue generation rather than cost avoidance: If the analytics work, the increase in revenue can be quite high relative to the cost of the system. A system that might cost a few thousand dollar could generate that in incremental sales in a month. By contrast, since security incidents are relatively rare, payback of surveillance analytics can be more difficult.
Intel has a whole program dedicated to the digital signage market. Additionally, Intel is advocating a digital signage interoperability specification. In an interview [link no longer available], Intel's digital signage director discussed in generalities how the video analytics could help adoption of Intel's archirecture in the market, mentioning January 2011 announcements for more information.
While Intel did not disclose the acquisition price, an industry tech site claims that the price was $17 Million for a company generating 'just under' $1 Million annual revenue. The site also mentions Quivodi as a much stronger direct competitor.
A number of video management/analytic companies have targeted both business/marketing applications as well as security. We think providing a simultaneously solution for both digital signage and surveillance will be difficult as little overlap in terms of technology, product or decision makers exist.