Video Annotation Based on Kernel Linear Neighborhood Propagation

Jinhui Tang, Xian-Sheng Hua, Guo-Jun Qi, Yan Song, Xiuqing Wu

Abstract

The insufficiency of labeled training data for representing the distribution of the entire dataset is a major obstacle in automatic semantic annotation of large-scale video database. Semi-supervised learning algorithms, which attempt to learn from both labeled and unlabeled data, are promising to solve this problem. In this paper, a novel graph-based semi-supervised learning method named <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">kernel</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">linear</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">neighborhood</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">propagation</i> (KLNP) is proposed and applied to video annotation. This approach combines the <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">consistency</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">assumption</i> , which is the basic assumption in semi-supervised learning, and the <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">local</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">linear</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">embedding</i> (LLE) method in a nonlinear kernel-mapped space. KLNP improves a recently proposed method <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">linear</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">neighborhood</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">propagation</i> (LNP) by tackling the limitation of its local linear assumption on the distribution of semantics. Experiments conducted on the TRECVID data set demonstrate that this approach outperforms other popular graph-based semi-supervised learning methods for video semantic annotation.
BibTeX
@article{tang2008video,
  author = {Jinhui Tang and Xian-Sheng Hua and Guo-Jun Qi and Yan Song and Xiuqing Wu},
  title = {{Video Annotation Based on Kernel Linear Neighborhood Propagation}},
  journal = {IEEE Trans. Multimedia},
  year = {2008},
  url = {https://openreview.net/forum?id=2i8oAhabzil},
}

Loading related content...

Figure