User Personalized Tag Based Image Search By Tag Relevance

Compuetr Engineering

With an increasing number of images that are available in social sites, image annotation has emerged as an important topic due to its application in image matching and retrieval. Most studies cast image annotation into a multi-label classification problem. Instead of having to make a binary decision for each tag, this approach ranks tags in the descending order of their relevance to the given image, significantly simplifying the problem. The proposed method aggregates the prediction models for different tags into a matrix, and casts tag ranking into a matrix recovery problem. It also contained user personalized view for searching images in a given image datasets. It introduces the matrix trace norm to control the model complexity, so that a reliable prediction model can learned for tag ranking even when the tag space is large and the number of training images is limited.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.