Discriminative dictionary learning via shared latent structure for object recognition and activity recognition

Hongcheng Wang, Hongbo Zhou, Alan Finn

Research output: Contribution to journalConference article

4 Citations (Scopus)

Abstract

We propose a novel low-dimensional discriminative dictionary learning approach for multi-class classification tasks, Latent Structure based Discriminative Dictionary Learning (LS-DDL). Our approach first projects features and class labels onto a shared latent structure space, and then generates a discriminative and low-dimensional input to a discriminative dictionary learning framework. LS-DDL learns a more discriminative and lower-dimensional dictionary than existing dictionary learning methods. Therefore we obtain high recognition accuracy with a small number of low-dimensional dictionary atoms. The low dimensionality also improves the efficiency in storage and testing. In addition, the latent structure projection eliminates the classifier weighting parameter in existing discriminative dictionary learning approaches. We validate the effectiveness and efficiency of the proposed approach through a series of experiments on image-based face recognition and video-based activity recognition. Our results show that the proposed approach obtains much higher recognition accuracy with a small number of dictionary atoms, and costs much less computational time than state-of-The-Art methods.

Original languageEnglish
Article number6907788
Pages (from-to)6299-6304
Number of pages6
JournalProceedings - IEEE International Conference on Robotics and Automation
DOIs
StatePublished - 22 Sep 2014
Event2014 IEEE International Conference on Robotics and Automation, ICRA 2014 - Hong Kong, China
Duration: 31 May 20147 Jun 2014

Fingerprint

Object recognition
Glossaries
Atoms
Face recognition
Labels
Classifiers
Testing

Cite this

@article{afe2e796d8fd4236bc2871a72116e3e5,
title = "Discriminative dictionary learning via shared latent structure for object recognition and activity recognition",
abstract = "We propose a novel low-dimensional discriminative dictionary learning approach for multi-class classification tasks, Latent Structure based Discriminative Dictionary Learning (LS-DDL). Our approach first projects features and class labels onto a shared latent structure space, and then generates a discriminative and low-dimensional input to a discriminative dictionary learning framework. LS-DDL learns a more discriminative and lower-dimensional dictionary than existing dictionary learning methods. Therefore we obtain high recognition accuracy with a small number of low-dimensional dictionary atoms. The low dimensionality also improves the efficiency in storage and testing. In addition, the latent structure projection eliminates the classifier weighting parameter in existing discriminative dictionary learning approaches. We validate the effectiveness and efficiency of the proposed approach through a series of experiments on image-based face recognition and video-based activity recognition. Our results show that the proposed approach obtains much higher recognition accuracy with a small number of dictionary atoms, and costs much less computational time than state-of-The-Art methods.",
author = "Hongcheng Wang and Hongbo Zhou and Alan Finn",
year = "2014",
month = "9",
day = "22",
doi = "10.1109/ICRA.2014.6907788",
language = "English",
pages = "6299--6304",
journal = "Proceedings - IEEE International Conference on Robotics and Automation",
issn = "1050-4729",

}

TY - JOUR

T1 - Discriminative dictionary learning via shared latent structure for object recognition and activity recognition

AU - Wang, Hongcheng

AU - Zhou, Hongbo

AU - Finn, Alan

PY - 2014/9/22

Y1 - 2014/9/22

N2 - We propose a novel low-dimensional discriminative dictionary learning approach for multi-class classification tasks, Latent Structure based Discriminative Dictionary Learning (LS-DDL). Our approach first projects features and class labels onto a shared latent structure space, and then generates a discriminative and low-dimensional input to a discriminative dictionary learning framework. LS-DDL learns a more discriminative and lower-dimensional dictionary than existing dictionary learning methods. Therefore we obtain high recognition accuracy with a small number of low-dimensional dictionary atoms. The low dimensionality also improves the efficiency in storage and testing. In addition, the latent structure projection eliminates the classifier weighting parameter in existing discriminative dictionary learning approaches. We validate the effectiveness and efficiency of the proposed approach through a series of experiments on image-based face recognition and video-based activity recognition. Our results show that the proposed approach obtains much higher recognition accuracy with a small number of dictionary atoms, and costs much less computational time than state-of-The-Art methods.

AB - We propose a novel low-dimensional discriminative dictionary learning approach for multi-class classification tasks, Latent Structure based Discriminative Dictionary Learning (LS-DDL). Our approach first projects features and class labels onto a shared latent structure space, and then generates a discriminative and low-dimensional input to a discriminative dictionary learning framework. LS-DDL learns a more discriminative and lower-dimensional dictionary than existing dictionary learning methods. Therefore we obtain high recognition accuracy with a small number of low-dimensional dictionary atoms. The low dimensionality also improves the efficiency in storage and testing. In addition, the latent structure projection eliminates the classifier weighting parameter in existing discriminative dictionary learning approaches. We validate the effectiveness and efficiency of the proposed approach through a series of experiments on image-based face recognition and video-based activity recognition. Our results show that the proposed approach obtains much higher recognition accuracy with a small number of dictionary atoms, and costs much less computational time than state-of-The-Art methods.

UR - http://www.scopus.com/inward/record.url?scp=84929192319&partnerID=8YFLogxK

U2 - 10.1109/ICRA.2014.6907788

DO - 10.1109/ICRA.2014.6907788

M3 - Conference article

AN - SCOPUS:84929192319

SP - 6299

EP - 6304

JO - Proceedings - IEEE International Conference on Robotics and Automation

JF - Proceedings - IEEE International Conference on Robotics and Automation

SN - 1050-4729

M1 - 6907788

ER -