DMCA
Locality-constrained linear coding for image classification (2010)
Cached
Download Links
| Venue: | IN: IEEE CONFERENCE ON COMPUTER VISION AND PATTERN CLASSIFICATOIN |
| Citations: | 442 - 20 self |
BibTeX
@INPROCEEDINGS{Wang10locality-constrainedlinear,
author = {Jinjun Wang and Jianchao Yang and Kai Yu and Fengjun Lv and Thomas Huang and Yihong Gong},
title = { Locality-constrained linear coding for image classification},
booktitle = {IN: IEEE CONFERENCE ON COMPUTER VISION AND PATTERN CLASSIFICATOIN},
year = {2010},
publisher = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
The traditional SPM approach based on bag-of-features (BoF) requires nonlinear classifiers to achieve good image classification performance. This paper presents a simple but effective coding scheme called Locality-constrained Linear Coding (LLC) in place of the VQ coding in traditional SPM. LLC utilizes the locality constraints to project each descriptor into its local-coordinate system, and the projected coordinates are integrated by max pooling to generate the final representation. With linear classifier, the proposed approach performs remarkably better than the traditional nonlinear SPM, achieving state-of-the-art performance on several benchmarks. Compared with the sparse coding strategy [22], the objective function used by LLC has an analytical solution. In addition, the paper proposes a fast approximated LLC method by first performing a K-nearest-neighbor search and then solving a constrained least square fitting problem, bearing computational complexity of O(M + K2). Hence even with very large codebooks, our system can still process multiple frames per second. This efficiency significantly adds to the practical values of LLC for real applications.
Keyphrases
locality-constrained linear coding image classification linear classifier large codebooks objective function real application effective coding scheme several benchmark final representation analytical solution traditional spm practical value traditional spm approach vq coding computational complexity nonlinear classifier multiple frame traditional nonlinear spm k-nearest-neighbor search square fitting problem state-of-the-art performance local-coordinate system good image classification performance fast approximated llc method locality constraint
