Smart Camera Advantages for Critical Security

Published Nov 07, 2010 00:00 AM
PUBLIC - This article does not require an IPVM subscription. Feel free to share.

In our video analytic testing series, we observed 2 key advantages for smart cameras (cameras with built-in analytics) vs cameras that have an add-on option or server based analytics. We think these advantages are important for demanding surveillance applications.

Traditionally, proponents for smart cameras argued the bandwidth saving benefits of smart cameras and the higher quality of uncompressed video. While there are certainly benefits of both, we think these are not the most important.

The two key advantages we found were:

  • Predictable performance from deployment to deployment: Because all elements are the same (camera, lens, encoder, etc.), the likelihood of surprise is low. You know what you are getting (good and bad).
  • Optimized for camera specifics: When analytics are built in to cameras, developers can optimize (or at least compensate) for camera specific issues.

Camera characteristics significantly impact the performance of video analytics. Of the top 4 most common issues we see (shadows, glare, low light noise and foliage), 3 of them significantly vary based on the properties of the camera.

Adding in analytics to 3rd party cameras expose the analytic to various video quality/issues. It's not possible to 'know' or adjust for the variety of video feeds that the analytics may need to process. We saw this repeatedly when using analytics with different cameras. Some cameras performed a lot better than others - mainly having to do with the shadow, glare and low light noise the camera displayed.

On the other hand, smart cameras are extremely restrictive. If you already have cameras in place (usually the case) or want to use a specific non-smart camera, using a smart camera can be difficult or unrealistic.

A big risk with analytic projects is cost-overruns and inability to get specific sites or cameras optimized. Smart cameras with built in analytics can help eliminate that. Once you test and understand the analytics strength and weakness, you can project where it will or will not work. Negative surprises are much less likely.

This being said, we still do not believe there is any cheap, easy way to guarantee video analytics do not generate significant false alerts. However, we do find cameras with built-in analytics as one means to reduce deployment issues.