Flexible top-view human pose estimation for detection system via CNN

Ryuji Go, Yoshimitsu Aoki

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    We propose the DeepPose-based pose estimation system that is flexible with the change of bounding-box range for top-view images. Our purpose is to link person detection system and pose estimation system. We introduce Bounding-box Curriculum Learning (BCL) and Recurrent Pose Estimation (RPE). BCL is a learning technique of CNN inspired from Curriculum Learning. RPE is a recurrent process of pose estimation that fixes the bounding-box range in response to the estimated results. We show the effect of proposed methods compared to normal learned CNN-based pose estimator on our original top-view dataset.

    Original languageEnglish
    Title of host publication2016 IEEE 5th Global Conference on Consumer Electronics, GCCE 2016
    PublisherInstitute of Electrical and Electronics Engineers Inc.
    ISBN (Electronic)9781509023332
    DOIs
    Publication statusPublished - 2016 Dec 27
    Event5th IEEE Global Conference on Consumer Electronics, GCCE 2016 - Kyoto, Japan
    Duration: 2016 Oct 112016 Oct 14

    Other

    Other5th IEEE Global Conference on Consumer Electronics, GCCE 2016
    CountryJapan
    CityKyoto
    Period16/10/1116/10/14

      Fingerprint

    Keywords

    • Convolutional Neural Networks
    • Pose Estimation
    • Top-view

    ASJC Scopus subject areas

    • Signal Processing
    • Electrical and Electronic Engineering
    • Computer Science Applications
    • Hardware and Architecture
    • Instrumentation

    Cite this

    Go, R., & Aoki, Y. (2016). Flexible top-view human pose estimation for detection system via CNN. In 2016 IEEE 5th Global Conference on Consumer Electronics, GCCE 2016 [7800406] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/GCCE.2016.7800406