Codebook quantization for image classification using incremental neural learning and subgraph extraction

Authors:
Ye Tang;Yu-Bin Yang;Yang Gao;Yao Zhang;Ying-Chun Cao
Affiliations:
State Key Lab for Novel Software Technology, Nanjing University, Nanjing, China;State Key Lab for Novel Software Technology, Nanjing University, Nanjing, China;State Key Lab for Novel Software Technology, Nanjing University, Nanjing, China;Jinling College, Nanjing University, Nanjing, China;State Key Lab for Novel Software Technology, Nanjing University, Nanjing, China
Venue:
IDEAL'12 Proceedings of the 13th international conference on Intelligent Data Engineering and Automated Learning
Year:
2012

Citing 7
Cited 0

Video Google: A Text Retrieval Approach to Object Matching in Videos

ICCV '03 Proceedings of the Ninth IEEE International Conference on Computer Vision - Volume 2
Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories

CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 2
An incremental network for on-line unsupervised classification and topology learning

Neural Networks
Towards optimal bag-of-features for object categorization and semantic video retrieval

Proceedings of the 6th ACM international conference on Image and video retrieval
Online dictionary learning for sparse coding

ICML '09 Proceedings of the 26th Annual International Conference on Machine Learning
Visual Word Ambiguity

IEEE Transactions on Pattern Analysis and Machine Intelligence
Exploring relations of visual codes for image classification

CVPR '11 Proceedings of the 2011 IEEE Conference on Computer Vision and Pattern Recognition

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper proposes a fast, incremental codebook quantization algorithm for image classification consisting of a fast codebook graph learning algorithm using incremental neural learning, and a subgraph-based coding method. Comparing with the algorithms based on classic Bag-of-Features (BOF) model, it holds the following advantages: 1) it learns codebook fast and effectively simply using a few training data; 2) it models relationships among visual words to guarantee higher discriminative power; 3) it automatically learns codebook with appropriate size. The above characteristics make our method more suitable for handling large-scale image classification tasks. Experimental results on Caltech-101 and Caltech-256 datasets demonstrate that the proposed algorithm achieves better performance while decreasing the computational cost remarkably.