Deep 3D pose dictionary: 3D human pose estimation from single RGB image using deep convolutional neural network

Reda Elbasiony, Walid Gomaa, Tetsuya Ogata

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

    Abstract

    In this work, we propose a new approach for 3D human pose estimation from a single monocular RGB image based on a deep convolutional neural network (CNN). The proposed method depends on reducing the huge search space of the continuous-valued 3D human poses by discretizing and approximating these continuous poses into many discrete key-poses. These key-poses constitute more restricted search space and then can be considered as multiple-class candidates of 3D human poses. Thus, a suitable classification technique is trained using a set of 3D key-poses and their corresponding RGB images to build a model to predict the 3D pose class of an input monocular RGB image. We use deep CNN as a suitable classifier because it is proven to be the most accurate technique for RGB image classification. Our approach is proven to achieve good accuracy which is comparable to the state-of-the-art methods.

    Original languageEnglish
    Title of host publicationArtificial Neural Networks and Machine Learning – ICANN 2018 - 27th International Conference on Artificial Neural Networks, 2018, Proceedings
    EditorsYannis Manolopoulos, Barbara Hammer, Vera Kurkova, Lazaros Iliadis, Ilias Maglogiannis
    PublisherSpringer-Verlag
    Pages310-320
    Number of pages11
    ISBN (Print)9783030014230
    DOIs
    Publication statusPublished - 2018 Jan 1
    Event27th International Conference on Artificial Neural Networks, ICANN 2018 - Rhodes, Greece
    Duration: 2018 Oct 42018 Oct 7

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume11141 LNCS
    ISSN (Print)0302-9743
    ISSN (Electronic)1611-3349

    Other

    Other27th International Conference on Artificial Neural Networks, ICANN 2018
    CountryGreece
    CityRhodes
    Period18/10/418/10/7

    Fingerprint

    Pose Estimation
    Glossaries
    Neural Networks
    Neural networks
    Search Space
    Image classification
    Classifiers
    Image Classification
    Classifier
    Predict
    Human
    Dictionary
    Class
    Model

    Keywords

    • 3D pose estimation
    • CNN
    • Deep learning
    • Human3.6m

    ASJC Scopus subject areas

    • Theoretical Computer Science
    • Computer Science(all)

    Cite this

    Elbasiony, R., Gomaa, W., & Ogata, T. (2018). Deep 3D pose dictionary: 3D human pose estimation from single RGB image using deep convolutional neural network. In Y. Manolopoulos, B. Hammer, V. Kurkova, L. Iliadis, & I. Maglogiannis (Eds.), Artificial Neural Networks and Machine Learning – ICANN 2018 - 27th International Conference on Artificial Neural Networks, 2018, Proceedings (pp. 310-320). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11141 LNCS). Springer-Verlag. https://doi.org/10.1007/978-3-030-01424-7_31

    Deep 3D pose dictionary : 3D human pose estimation from single RGB image using deep convolutional neural network. / Elbasiony, Reda; Gomaa, Walid; Ogata, Tetsuya.

    Artificial Neural Networks and Machine Learning – ICANN 2018 - 27th International Conference on Artificial Neural Networks, 2018, Proceedings. ed. / Yannis Manolopoulos; Barbara Hammer; Vera Kurkova; Lazaros Iliadis; Ilias Maglogiannis. Springer-Verlag, 2018. p. 310-320 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11141 LNCS).

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

    Elbasiony, R, Gomaa, W & Ogata, T 2018, Deep 3D pose dictionary: 3D human pose estimation from single RGB image using deep convolutional neural network. in Y Manolopoulos, B Hammer, V Kurkova, L Iliadis & I Maglogiannis (eds), Artificial Neural Networks and Machine Learning – ICANN 2018 - 27th International Conference on Artificial Neural Networks, 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11141 LNCS, Springer-Verlag, pp. 310-320, 27th International Conference on Artificial Neural Networks, ICANN 2018, Rhodes, Greece, 18/10/4. https://doi.org/10.1007/978-3-030-01424-7_31
    Elbasiony R, Gomaa W, Ogata T. Deep 3D pose dictionary: 3D human pose estimation from single RGB image using deep convolutional neural network. In Manolopoulos Y, Hammer B, Kurkova V, Iliadis L, Maglogiannis I, editors, Artificial Neural Networks and Machine Learning – ICANN 2018 - 27th International Conference on Artificial Neural Networks, 2018, Proceedings. Springer-Verlag. 2018. p. 310-320. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-01424-7_31
    Elbasiony, Reda ; Gomaa, Walid ; Ogata, Tetsuya. / Deep 3D pose dictionary : 3D human pose estimation from single RGB image using deep convolutional neural network. Artificial Neural Networks and Machine Learning – ICANN 2018 - 27th International Conference on Artificial Neural Networks, 2018, Proceedings. editor / Yannis Manolopoulos ; Barbara Hammer ; Vera Kurkova ; Lazaros Iliadis ; Ilias Maglogiannis. Springer-Verlag, 2018. pp. 310-320 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
    @inproceedings{3375f0f12f054bd78ddfbd5c86fc21e2,
    title = "Deep 3D pose dictionary: 3D human pose estimation from single RGB image using deep convolutional neural network",
    abstract = "In this work, we propose a new approach for 3D human pose estimation from a single monocular RGB image based on a deep convolutional neural network (CNN). The proposed method depends on reducing the huge search space of the continuous-valued 3D human poses by discretizing and approximating these continuous poses into many discrete key-poses. These key-poses constitute more restricted search space and then can be considered as multiple-class candidates of 3D human poses. Thus, a suitable classification technique is trained using a set of 3D key-poses and their corresponding RGB images to build a model to predict the 3D pose class of an input monocular RGB image. We use deep CNN as a suitable classifier because it is proven to be the most accurate technique for RGB image classification. Our approach is proven to achieve good accuracy which is comparable to the state-of-the-art methods.",
    keywords = "3D pose estimation, CNN, Deep learning, Human3.6m",
    author = "Reda Elbasiony and Walid Gomaa and Tetsuya Ogata",
    year = "2018",
    month = "1",
    day = "1",
    doi = "10.1007/978-3-030-01424-7_31",
    language = "English",
    isbn = "9783030014230",
    series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
    publisher = "Springer-Verlag",
    pages = "310--320",
    editor = "Yannis Manolopoulos and Barbara Hammer and Vera Kurkova and Lazaros Iliadis and Ilias Maglogiannis",
    booktitle = "Artificial Neural Networks and Machine Learning – ICANN 2018 - 27th International Conference on Artificial Neural Networks, 2018, Proceedings",

    }

    TY - GEN

    T1 - Deep 3D pose dictionary

    T2 - 3D human pose estimation from single RGB image using deep convolutional neural network

    AU - Elbasiony, Reda

    AU - Gomaa, Walid

    AU - Ogata, Tetsuya

    PY - 2018/1/1

    Y1 - 2018/1/1

    N2 - In this work, we propose a new approach for 3D human pose estimation from a single monocular RGB image based on a deep convolutional neural network (CNN). The proposed method depends on reducing the huge search space of the continuous-valued 3D human poses by discretizing and approximating these continuous poses into many discrete key-poses. These key-poses constitute more restricted search space and then can be considered as multiple-class candidates of 3D human poses. Thus, a suitable classification technique is trained using a set of 3D key-poses and their corresponding RGB images to build a model to predict the 3D pose class of an input monocular RGB image. We use deep CNN as a suitable classifier because it is proven to be the most accurate technique for RGB image classification. Our approach is proven to achieve good accuracy which is comparable to the state-of-the-art methods.

    AB - In this work, we propose a new approach for 3D human pose estimation from a single monocular RGB image based on a deep convolutional neural network (CNN). The proposed method depends on reducing the huge search space of the continuous-valued 3D human poses by discretizing and approximating these continuous poses into many discrete key-poses. These key-poses constitute more restricted search space and then can be considered as multiple-class candidates of 3D human poses. Thus, a suitable classification technique is trained using a set of 3D key-poses and their corresponding RGB images to build a model to predict the 3D pose class of an input monocular RGB image. We use deep CNN as a suitable classifier because it is proven to be the most accurate technique for RGB image classification. Our approach is proven to achieve good accuracy which is comparable to the state-of-the-art methods.

    KW - 3D pose estimation

    KW - CNN

    KW - Deep learning

    KW - Human3.6m

    UR - http://www.scopus.com/inward/record.url?scp=85054815546&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85054815546&partnerID=8YFLogxK

    U2 - 10.1007/978-3-030-01424-7_31

    DO - 10.1007/978-3-030-01424-7_31

    M3 - Conference contribution

    AN - SCOPUS:85054815546

    SN - 9783030014230

    T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    SP - 310

    EP - 320

    BT - Artificial Neural Networks and Machine Learning – ICANN 2018 - 27th International Conference on Artificial Neural Networks, 2018, Proceedings

    A2 - Manolopoulos, Yannis

    A2 - Hammer, Barbara

    A2 - Kurkova, Vera

    A2 - Iliadis, Lazaros

    A2 - Maglogiannis, Ilias

    PB - Springer-Verlag

    ER -