Challenges with H.264 Streams at the CoreBy John Honovich, Published on Aug 26, 2009
In the debate of edge vs core video analytics, one of the key issues and drivers towards the edge is the challenges procesing H.264 video streams at the core. Specifically, H.264 streams processed at the core can reduce accuracy of the analytics and significantly increase the processing load on the server that is decoding the H.264 to perform the analysis. By contrast, at the edge, uncompressed video can be immeditely anlayzed. This is similar to the practice used by 'smart' DVRs to analyze feeds from analog cameras.
The overhead and issues with processing H.264 streams at the core is a major reason why NVR providers are offering or integrating with 3rd party IP cameras. 3VR's Director of Engineering notes this problem in a recent analysis of technical aspects of using IP cameras. While 3VR is primarily a DVR/NVR manufacturer, last year they introduced a line of smart cameras that would help ameloriate this problem (though at the expense of requiring the use of their smart cameras rather than normal COTS IP cameras).