LAS VEGAS — During Thursday’s “Video Analytics and Content Analysis: Use Cases Explored, RFPs Developed” panel at ISC West, attendees learned about the latest video analytics solutions and best practices for writing request for proposals (RFPs) to deploy the technology.
SSI Editor-in-Chief Scott Goldfine moderated the session, which was part of the ISC Premier Education Series, Video Technology track. The four-person panel of video analytics experts included: Zvika Ashani, chief technology officer and co-founder, Agent Vi; Bob Cutting, vice president of product management, ObjectVideo; Amit Gavish, general manager for North America, Briefcam; and Malay Kundu, founder and CEO, StopLift Checkout Vision Systems. Each panel member shared a case study explaining the different uses for video analytics and video synopsis technology to a room full of integrators and end users.
Kundu opened with a demonstration on how deploying video analytics effectively in retail applications reduces shrink. His company offers ScanItAll, a solution that automatically analyzes CCTV video to detect various forms of theft, training error and operational analytics at the checkout counter.
As a result of constant inventory shrink, the solution was installed at 60 Big Y Foods chain stores. Attendees viewed footage of Big Y employees and customers avoiding scanning items at checkout. “Big Y has a typical profile of retailers here in the U.S.,” said Kundu. “For every $100 of inventory shrinkage, roughly $50 of that is lost to employee theft. Out of that $50, $35 is lost at the checkout. Out of that, $25 to $30 is lost specifically because of scan avoidance.”
To combat this problem at any retail application, Kundu suggested installing IP or analog cameras over each lane, while using video analytics technology.
Cutting followed next with his case study of the Chumash Casino Resort in Santa Ynez, Calif., where people counting and tracking floor traffic patterns proved to be an issue. Because physical sensors were not consistent with the casino’s design, employees often served as people counters. Unfortunately, that staff proved unreliable. Thus, by relying on guess trends and not real-time data, the casino overstaffed employees because it was unclear on how many workers were needed each day.
Chumash installed ObjectVideo’s video analytics solution at the entrances and exits of the casino. As a result, the end user saw an improvement in employee management. Additionally, the casino integrated the solution into their slot machine system to evaluate how many people were using slot machines on the casino floor.
“The beauty of this deployment is its simplicity,” Cutting said. “Chumash is one of ObjectVideo’s smaller clients. This proves that video analytics does not have to be a big, complex system. With clear business drivers and direction, and a focused use of video analytics, you can have high video analytic performance. You can also deliver a significant ROI with the customer.”
Next up was Ashani, who demonstrated how video analytics can be effective for people counting in the transportation market. “In order for video analytics algorithms to perform well and provide the levels that are expected from the end users, it really needs to be tailored for specific applications,” he said.
Ashani explained that companies operating a fleet of buses or trains need to gather information on how buses operate. Thus, by using video analytics, operators can determine how many people enter or exit a bus/train at each stop, and how many people are on it while in motion. “We’re really gathering information in order to perform optimization of operations,” Ashani said.
He added that deploying video analytics on a moving platform can be challenging because the scene is constantly changing, coupled with a narrow field of view and drastic lighting conditions, so algorithms need to be adjusted accordingly.
In the final demonstration, Gavish discussed video synopsis, an analytical tool that allows many hours of activities to be reviewed in minutes. “We’re able to add more and more cameras to a system over time,” he said. “Now we need to address how much video can be reviewed in a short amount of time.
He likened video synopsis to a book, which features an executive summary and an index. “The executive summary quickly reviews everything you’re about to read in the book,” Gavish said. “The index helps readers find where so-and-so was mentioned in the book. Both make it easier to find events in the content.”
The technology combines all the events that occurred in a day into one video. If a user sees something suspicious in the footage, he or she can click on the person in question, and view the original footage, said Gavish.
Later, each panelist offered advice for integrators who are planning video analytics deployments. Following are a few tips from the experts:
- Clearly define the customer’s analytics requirements
- Verify which requirement is achievable by the solution
- Determine appropriate camera types to uses, including camera placement and field of view
- Analyze lighting requirements
- Determine the analytics rules that will be applied to each camera
- Define where and how the analytics results will be viewed
- Determine the required server (hardware or software)
- Define clear and reasonable acceptance criteria
“Don’t make assumptions in what video analytics solutions will solve the problem,” said Cutting. “Early in the process, you may not know. What I want to see is how you intend for the solution to give you what you want. There needs to be some application for what you want to accomplish at that business level. It helps the analytics vendor be able to come back and respond to you.”
Ashley Willis is associate editor for SECURITY SALES & INTEGRATION.