Person re-id by incorporating PCA loss in CNN

  • Kaixuan Zhang
  • , Yang Xu
  • , Li Sun*
  • , Song Qiu
  • , Qingli Li
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This paper proposes an algorithm, particularly a loss function and its end to end learning manner, for person re-identification task. The main idea is to take full advantage of the labels in a batch during training, and to employ PCA to extract discriminative features. Deriving from the classic eigenvalue computation problem in PCA, our method incorporates an extra term in loss function with the purpose of minimizing those relative large eigenvalues. And the derivative with respect to the designed loss can be back-propagated in deep network by stochastic gradient descent (SGD). Experiments show the effectiveness of our algorithm on several re-id datasets.

Original languageEnglish
Title of host publicationMultiMedia Modeling - 24th International Conference, MMM 2018, Proceedings
EditorsAhmed Elgammal, Thanarat H. Chalidabhongse, Supavadee Aramvith, Yo-Sung Ho, Klaus Schoeffmann, Chong Wah Ngo, Noel E. O'Connor, Moncef Gabbouj
PublisherSpringer Verlag
Pages200-212
Number of pages13
ISBN (Print)9783319735993
DOIs
StatePublished - 2018
Event24th International Conference on MultiMedia Modeling, MMM 2018 - Bangkok, Thailand
Duration: 5 Feb 20187 Feb 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10705 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference24th International Conference on MultiMedia Modeling, MMM 2018
Country/TerritoryThailand
CityBangkok
Period5/02/187/02/18

Fingerprint

Dive into the research topics of 'Person re-id by incorporating PCA loss in CNN'. Together they form a unique fingerprint.

Cite this