Large-scale image/videos´ automatic annotation and retrieval based on the distributed users

In this paper, a mechanism of large-scale image/videos (I/Vs) automatic annotation and retrieval based on the distributed users is proposed, and a prototype system on the basis of it is implemented by the .net framework. Furthermore, the algorithm of annotating the meanings of I/Vs by their associative values with predefined categories, and the method of implementing the system is described. Experiment results regarding the evaluation that users can find their expecting I/Vs show that the proposed approach is effective and efficient for annotating and retrieving large I/Vs.