Image Annotation Incorporating Low-Rankness, Tag and Visual Correlation and Inhomogeneous Errors
This work addresses tag-based image management applications by improving annotation accuracy, though it appears incremental as it builds on existing priors and methods.
The paper tackles the problem of incomplete and inaccurate tags in tag-based image retrieval by proposing a novel image annotation method that incorporates low-rankness, tag and visual correlation, and inhomogeneous errors, using CNN features and word vectors, and demonstrates effectiveness and robustness on multiple benchmark datasets.
Tag-based image retrieval (TBIR) has drawn much attention in recent years due to the explosive amount of digital images and crowdsourcing tags. However, TBIR is still suffering from the incomplete and inaccurate tags provided by users, posing a great challenge for tag-based image management applications. In this work, we proposed a novel method for image annotation, incorporating several priors: Low-Rankness, Tag and Visual Correlation and Inhomogeneous Errors. Highly representative CNN feature vectors are adopt to model the tag-visual correlation and narrow the semantic gap. And we extract word vectors for tags to measure similarity between tags in the semantic level, which is more accurate than traditional frequency-based or graph-based methods. We utilize the accelerated proximal gradient (APG) method to solve our model efficiently. Extensive experiments conducted on multiple benchmark datasets demonstrate the effectiveness and robustness of the proposed method.