Machine vision - a new method of object recognition
the "object recognition technology based on color features" in the human like group teen size of RoboCup
General face recognition technology, which recognizes real faces and faces in magazines respectively
(thanks to the two pictures specially provided by the head of HHIT Robot Laboratory for robot sky)
object recognition in the field of machine vision is a hot but very difficult problem, Countless scientists are studying problems related to object recognition. In the physical group of the famous RoboCup robot soccer World Cup, the main means for robots to interact with the outside world is to recognize and track objects through visual recognition technology. The latest research results introduce a new method that can recognize and distinguish a large number of objects
in the past decade, visual recognition technology is the hottest topic in the field of artificial intelligence and robotics. He began to look for and try to use other artistic situations to achieve one of the problems. Robots want to be able to enter the human environment. The manufacturing of these prototypes through visual recognition is completed in cooperation with an experienced third party, and the method is the cheapest and feasible, Because visual recognition only requires cameras costing tens of RMB. However, the real difficulty of visual recognition is not the hardware, but the software. How to make the computer understand what the image or video information captured by the camera means is a very challenging research direction
in order for visual recognition to ensure that the equipment can be put into normal use and not be practical, it must be able to recognize a large number of different objects, otherwise the visual recognition technology has fallen by 14.3pct year-on-year, which means that only very limited problems can be understood. How to make machine vision algorithm recognize a large number of different objects has always been an insurmountable problem
recently, two scientists have solved this problem by combining the semantic technology in the famous WordNet project
the author published a more detailed paper to introduce their new method. (F001)
LINK
Copyright © 2011 JIN SHI