Online photo services such as Flickr allow users to share their photos with family, friends, and the online community at large. Users manually annotate their photos using so called tags, which describe the contents of the photo. These user-provided tags on images have been demonstrated crucial for search. However, the recent research shows that queries about people's opinions occur quite often, while people tend to neglect this category for tagging. In this paper, we proposed a novel framework to automatically annotate opinion tags to images. It can automatically find people's most representative opinions on a given image from the Internet by near duplicate image detection and opinion mining technologies. Experimental results demonstrate the effectiveness of the proposed method. |