Subspace Clustering Based Tag Sharing for Inductive Tag Matrix Refinement with Complex Errors

TL;DR

This paper introduces a new image annotation framework that combines subspace clustering and a matrix completion model to improve tag accuracy and completeness, especially in noisy datasets.

Contribution

It presents a novel combination of sparse subspace clustering and a matrix completion model that accounts for complex errors in image annotation.

Findings

Outperforms state-of-the-art methods on benchmark datasets.

Effectively handles annotation noise and errors.

Improves tag completion and refinement accuracy.

Abstract

Annotating images with tags is useful for indexing and retrieving images. However, many available annotation data include missing or inaccurate annotations. In this paper, we propose an image annotation framework which sequentially performs tag completion and refinement. We utilize the subspace property of data via sparse subspace clustering for tag completion. Then we propose a novel matrix completion model for tag refinement, integrating visual correlation, semantic correlation and the novelly studied property of complex errors. The proposed method outperforms the state-of-the-art approaches on multiple benchmark datasets even when they contain certain levels of annotation noise.