Intelligent VideoSmart cameras will redefine what makes useful dataMost people hearing the words "intelligent video" are likely to think about PBS or the History Channel. Technologies and executive attitudes have started to coalesce, however, so that that perspective is beginning to change. Smile, You're on Candid CameraConsider a possible CRM application: A customer walks into a sports store and has her face recognized by a system that identifies her as someone who occasionally buys tennis equipment. As she moves past the entrance, an electronic bulletin board announces sales on tennis rackets. Meanwhile, a camera in the parking lot notifies a clerk in the store that the customer's car is at the far end of the parking lot, so she'll need help carrying packages. The camera also notices a bike rack on the back of her car, which triggers another announcement as the customer leaves about a charity race over the weekend, and records the bicycle information for later use. In fact, these cameras are so into the game that they talk to each other as the customer moves away from her car after loading it and heads to another store in the mall. A hint about her related purchases or recreational timeouts from a shopping errand has economic value — so after the customer leaves its line of vision, one camera tells another to note the stores she visits or musicians she listens to. Smart CamerasWhat makes this system technologically feasible (putting aside the privacy and policy issues for a moment) is the arrival of the smart camera, which contains a microprocessor that does what a team of humans would do if you could afford to pay a zillion of them to monitor, record, interpret, archive, retrieve, verify, and communicate the image data effectively. Dumb cameras require too many eyeballs to process all the information they can access. Early smart cameras were PC-based, leading to problems with installation and maintenance, and tracked pixels instead of objects. Consequently, they were easily fooled by changes in lighting or movement. Facial recognition software often produced unacceptable false alarms, identifying someone incorrectly. Wind or even vibrations from air conditioners could jerk around cameras so that images often had little utility. Enabling TechnologiesThese problems are being overcome with video processors — microprocessors that increase performance, improve reliability, and reduce costs, and are especially designed to process video data. A leader in the field is Equator Technologies. (Full disclosure: I am an advisor for the company.) Its programmable systems-on-a-chip solutions can replace the hardwired chips frequently used in the past to satisfy the huge computational and performance requirements for video compression and analysis. The adaptive nature of the solution welcomes smarter and more innovative software applications. Thus, video intelligence can move more quickly from basic vision functions, such as compression and motion tracking, to more sophisticated "interpretive vision" (a term coined at Sarnoff Corp.) that includes pattern analysis and activity monitoring. MindsetsAs bandwidth increases, wireless technologies become ubiquitous, and digital camera costs decline, the technological obstacles to widespread networked solutions dissipate. Equally important, business is beginning to recognize camera deployment as more than just a cost associated with security (documenting theft and providing deterrence) and more as a tool for generating revenue. Indeed, some have argued that consumer applications, such as video mail, represent the next big thing. Imagine a working mom getting a message on her office PC that says, "Hey, you might want to look at this," just as Grandma goes for a nap while leaving the gas oven on in the kitchen. The smart camera sends an alert when something important is happening vs. the nanny cam, which gives access to hours and hours of useless "nothing is happening" monitoring. Nonetheless, business-to-business applications abound. For example, a retailer could use squads of observers to watch how people move around a store, noting the sequence of counters visited, time spent at different locations, the displays that received attention, or the shelves or racks preferred by the customer. Or they could do that with cameras. Peter DiAngelis, President and CEO of camera manufacturer IQinVision, says, "The firms focused on store layouts need the best data to answer their hardest, big payoff questions. They could care less whether it comes from humans on the ground or cameras that imitate them. They just see the money and customer satisfaction that come from smarter video intelligence." At the least, measuring the effectiveness of in-store advertising would improve; indeed, it could be monetized in the manner of co-advertising programs. Privacy and Village LifeAlthough people don't have to be identified, only observed or seen, with these and other types of multitracking situations, privacy advocates are unnerved. The operating metaphor is increasingly about life in a village — where everyone has the capacity to know who you are and your patterns of behavior. If it's true that technological advances can't be restrained, then it's also true that business will find ways to use those advances in the pursuit of profit. Business leaders would be wise to strategize about the policy implications, not to mention the revenue-producing prospects, of a business opportunity that is now gathering momentum. Nicholas Imparato [imparato@hoover.stanford.edu] is a research fellow at the Hoover Institution, a professor of management and marketing at the University of San Francisco, the editor of Public Policy and the Internet (Hoover Press, 2000), and a director/advisory board member of several firms in the IT sector.
|
 |
Most Popular This Week
IE Weekly Newsletter
Subscribe to the newsletter
| |
|
|