Please note that JavaScript and style sheet are used in this website,
Due to unadaptability of the style sheet with the browser used in your computer, pages may not look as original.
Even in such a case, however, the contents can be used safely.
Video-image recognition and understanding
Video and Image Understanding technology
In order to realize a highly efficient and safe society with IT, technologies that can accurately sense and understand the real world are inevitable. In particular, image and video analyzing technologies that can replace human vision are becoming increasingly important. Built upon a long history of intensive research on image processing, pattern recognition and machine learning technologies, the Information and Media Processing Laboratories is now one of the global leaders in image recognition and real world sensing technologies, including face recognition, human behavior analysis and object recognition.
Face Recognition Technology
In face recognition, the verification image and the registered image to be compared may be captured under different conditions, e.g. with different postures, lightings, expressions, ages, etc. This can substantially degrade the verification performance. NEC has developed two key-technologies: a perturbation space method, which generates images with different postures and lighting conditions from a single image, and GLVQ-based Multiple-Matching Face Detector, which extracts useful features for identifying individuals. These two technologies enable robust and high-accuracy face identification under diverse conditions.
NEC’s face recognition technology has won the 1st place in the global competition on face recognition*, and is acclaimed as the world’s leading technology in the field.
* http://www.nist.gov/itl/iad/ig/mbe.cfm
Human Behavior Analysis System
Human behavior analysis is a technology for detecting, extracting, tracking and analyzing humans in videos and images. This technology allows instant discovery and notification of anomalies in surveillance footages. By analyzing the flow of people in stores or plants, this technology can be also used for enhancing sales and improving work efficiency. We are also working on the development of video search technology that incorporates this technology, as well as improving the robustness against variable lighting conditions.
Object Recognition Technology
Object recognition is a technology for extracting and recognizing regions of interest, such as objects and characters, from an image. The Information and Media Processing Laboratories aim to solve the difficult technical challenge of accurate and fast identification and classification of all kinds of objects through proprietary machine learning methods. We are also working on designing identification algorithms and developing technologies for efficient learning of numerous items.
Through research and development of these image recognition technologies, NEC will create physical security and next-generation video utilization services, and contribute to the creation of safe, secure and affluent society.
