Relevance feedback based on active learning and GMM in image retrieval system
Shuo Wang, Jianjian Wang · 2014
The image annotation and retrieval are significant for semantic image retrieval that needs to establish the relations between linguistic labels and images. So the probabilistic formulation for semantic labeling is introduced to solve them. In addition, relevance feedback can improve the retrieval performance efficiently in the content-based image retrieval (CBIR). In this paper, we proposed a new feedback approach with active learning method combined with Gaussian Mixture Model (GMM) which is used for the likelihood computation for the linguistic indexing.