3 ms·
This general concept has existed for a while, and the monitoring is often done in the US. The basic premise is to use video analytics to spot anomalies, most co
by brk 5y ago
This general concept has existed for a while, and the monitoring is often done in the US. The basic premise is to use video analytics to spot anomalies, most commonly along the lines of people jumping a fence to a secured property, and then sending an alert to the central station for an operator to review and act on.
In recent years, video analytics have (sort of) evolved to the point that some degree of gesture and action recognition can be implemented to spot acts of crime or aggression.
While the person-jumping-fence scenario can be repeatedly detected with a high degree of confidence, and false alarm scenarios like a cat or raccoon climbing the fence ignored, action/gesture recognition is far less accurate overall. I think this is why they are outsourcing monitoring to India instead of using one of the many domestic video monitoring operations (eg: Rapid Response, COPS, etc.). I would wager that maybe 1 in 50 of the events the remote operator gets alerted to is actually something of interest.
The remote audio talk down is very effective though, although these systems are more commonly used for outdoor theft and vandalism protection. The scenarios here of an indoor robbery are somewhat extreme. Typically if a store is at THAT much risk for robbery they end up installing bullet-proof plexi or taking other steps that are more effective in reducing risk than a remote operator and a speaker.
Source: this general market and application has been my focus for the last decade+