NEC Corporation has developed and tested large scale video processing technologies that can process advanced video analysis such as facial recognition and human identification three times more efficiently than conventional technologies on a single server.
These technologies were tested on video surveillance systems covering large areas, which analyze information on peoples’ characteristics,...
NEC Corporation has developed and tested large scale video processing technologies that can process advanced video analysis such as facial recognition and human identification three times more efficiently than conventional technologies on a single server.
These technologies were tested on video surveillance systems covering large areas, which analyze information on peoples' characteristics, including facial features, clothing features and suspicious behaviors. In these tests, NEC's new technologies enabled a single server (*) to completely analyze images from thirty surveillance cameras, three times more cameras than a single server can process with existing technologies. As a result, a company can utilize its existing servers to monitor approximately three times more area by adopting NEC's advanced image analysis technologies.
In recent years, there is growing demand worldwide for automatic video analysis of surveillance camera content from areas that include airports, train stations and city streets. Conventionally, in order to meet the advanced processing requirements of analyzing video data from multiple cameras, companies needed to maintain numerous servers. This costly demand for hardware could force companies to restrain expenses by limiting their automatic video analysis to only the most important areas, thereby leaving large areas underserved by surveillance.
“NEC's technologies meet video processing demand by controlling the level of processing in response to the content of images captured by surveillance cameras,” said Yasunori Mochizuki, Vice President, Information and Media Processing Laboratories, NEC. “For example, specialized controls can streamline processing efficiency by only analyzing images where people appear or by reducing the frequency of analysis for areas with a low level of priority.”
Key features of NEC's newly developed video surveillance technologies include the following:
1. Streamlined processing though analysis triggered by image content
NEC developed image processing control technologies that maintain processing accuracy and adjust video analysis based on its content. In this way, image processing frequency can be customized to focus on video featuring human subjects. Video without people can be analyzed at a lower frequency. These technologies can be further streamlined by adjusting the frequency of analysis in accordance with factors such as the time of day and important locations. This maximizes the amount of processing per server and expands the total area that can be covered by advanced image analysis using existing servers.
2. Streamlined processing in accordance with user preferences
Image processing control can be configured to meet user preferences, such as minimizing the amount of video that is analyzed and designating the most important areas for rapid image analysis.
These newly developed surveillance technologies, featuring NEC's highly regarded face recognition and human identification technologies, can potentially be deployed in countries throughout the world in order to improve the range and security of monitored areas.
In addition to video surveillance, NEC will continue to drive the research and development of big data processing technologies. Furthermore, the company will actively participate in trials domestically and abroad that help to quickly deliver its technologies to consumers anywhere.
NEC announces test results using these video surveillance processing technologies at the Forum on Information Technology 2012 (FIT2012), held at Hosei University's Koganei campus from 4 - 6 September 2012.