The ability to simultaneously monitor, process and understand semantic content from multiple video streams–broadcast over separate channels or transmitted via the Internet–is a capability that has long presented numerous technical challenges. As a result, this has prevented providers from being able to effectively allocate bandwidth to reduce transmission loads and ensure that certain video packets are transmitted at a faster speed based on their classification.
This invention describes a large-scale video stream monitoring system that provides the ability to gather and analyze video packets from multiple sources in real-time, and detect, process and understand the nature of the content being transmitted. After analyzing and processing the data, the monitoring system uses the results to filter the video packets and route the content in way that delivers better performance.
IBM inventors Lisa Amini of Yorktown Heights, NY; Ching-Yung Lin of Forest Hills, NY; and Olivier Verscheure of Hopewell Junction, NY created the patent.