Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis

Zhao Wang, Jun Ohya

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

    2 Citations (Scopus)

    Abstract

    This paper proposes a temporal Grouping and pattern analysis-based algorithm that could track the fingertips of guitarists during their guitar playing towards the actualization of the automatic guitar fingering recognition system. First a machine learning-based Bayesian Pixel Classifier is used to segment the hand area on the test data. Then, the probability map of fingertip is generated on the segmentation results by counting the voting numbers of the Template Matching and Reversed Hough Transform. Furthermore, a temporal Grouping algorithm, which is a geometry analysis for consecutive frames, is applied to removal noise and group the same fingertips (index finger, middle finger, ring finger, little finger). Then, a data association algorithm is utilized to associate 4 tracked fingers (index finger, middle finger, ring finger, little finger) with their correspondent tracked results frame by frame. Finally, particles are distributed only between the associated fingertip candidates to track the fingertips of guitarist effectively. The experimental result demonstrates that this fingertip tracking algorithm is robust enough for tracking fingertips (1) without any constrains such us color marker; (2) under the complex contexts, such us complicated background, different illumination conditions, (3) with the high tracking accuracy (mean error 3.36 pixels for four fingertips).

    Original languageEnglish
    Title of host publicationComputer Vision - ACCV 2016 Workshops, ACCV 2016 International Workshops, Revised Selected Papers
    PublisherSpringer Verlag
    Pages212-226
    Number of pages15
    Volume10118 LNCS
    ISBN (Print)9783319545257
    DOIs
    Publication statusPublished - 2017
    Event13th Asian Conference on Computer Vision, ACCV 2016 - Taipei, Taiwan, Province of China
    Duration: 2016 Nov 202016 Nov 24

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume10118 LNCS
    ISSN (Print)03029743
    ISSN (Electronic)16113349

    Other

    Other13th Asian Conference on Computer Vision, ACCV 2016
    CountryTaiwan, Province of China
    City Taipei
    Period16/11/2016/11/24

    Fingerprint

    Pattern Analysis
    Grouping
    Pixel
    Pixels
    Ring
    Noise Removal
    Data Association
    Template matching
    Hough Transform
    Template Matching
    Hough transforms
    Voting
    Learning systems
    Consecutive
    Illumination
    Counting
    Machine Learning
    Classifiers
    Segmentation
    Lighting

    ASJC Scopus subject areas

    • Theoretical Computer Science
    • Computer Science(all)

    Cite this

    Wang, Z., & Ohya, J. (2017). Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis. In Computer Vision - ACCV 2016 Workshops, ACCV 2016 International Workshops, Revised Selected Papers (Vol. 10118 LNCS, pp. 212-226). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10118 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-54526-4_16

    Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis. / Wang, Zhao; Ohya, Jun.

    Computer Vision - ACCV 2016 Workshops, ACCV 2016 International Workshops, Revised Selected Papers. Vol. 10118 LNCS Springer Verlag, 2017. p. 212-226 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10118 LNCS).

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

    Wang, Z & Ohya, J 2017, Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis. in Computer Vision - ACCV 2016 Workshops, ACCV 2016 International Workshops, Revised Selected Papers. vol. 10118 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10118 LNCS, Springer Verlag, pp. 212-226, 13th Asian Conference on Computer Vision, ACCV 2016, Taipei, Taiwan, Province of China, 16/11/20. https://doi.org/10.1007/978-3-319-54526-4_16
    Wang Z, Ohya J. Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis. In Computer Vision - ACCV 2016 Workshops, ACCV 2016 International Workshops, Revised Selected Papers. Vol. 10118 LNCS. Springer Verlag. 2017. p. 212-226. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-54526-4_16
    Wang, Zhao ; Ohya, Jun. / Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis. Computer Vision - ACCV 2016 Workshops, ACCV 2016 International Workshops, Revised Selected Papers. Vol. 10118 LNCS Springer Verlag, 2017. pp. 212-226 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
    @inproceedings{8331e490ceda44798c2b67c5bc96c62f,
    title = "Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis",
    abstract = "This paper proposes a temporal Grouping and pattern analysis-based algorithm that could track the fingertips of guitarists during their guitar playing towards the actualization of the automatic guitar fingering recognition system. First a machine learning-based Bayesian Pixel Classifier is used to segment the hand area on the test data. Then, the probability map of fingertip is generated on the segmentation results by counting the voting numbers of the Template Matching and Reversed Hough Transform. Furthermore, a temporal Grouping algorithm, which is a geometry analysis for consecutive frames, is applied to removal noise and group the same fingertips (index finger, middle finger, ring finger, little finger). Then, a data association algorithm is utilized to associate 4 tracked fingers (index finger, middle finger, ring finger, little finger) with their correspondent tracked results frame by frame. Finally, particles are distributed only between the associated fingertip candidates to track the fingertips of guitarist effectively. The experimental result demonstrates that this fingertip tracking algorithm is robust enough for tracking fingertips (1) without any constrains such us color marker; (2) under the complex contexts, such us complicated background, different illumination conditions, (3) with the high tracking accuracy (mean error 3.36 pixels for four fingertips).",
    author = "Zhao Wang and Jun Ohya",
    year = "2017",
    doi = "10.1007/978-3-319-54526-4_16",
    language = "English",
    isbn = "9783319545257",
    volume = "10118 LNCS",
    series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
    publisher = "Springer Verlag",
    pages = "212--226",
    booktitle = "Computer Vision - ACCV 2016 Workshops, ACCV 2016 International Workshops, Revised Selected Papers",
    address = "Germany",

    }

    TY - GEN

    T1 - Fingertips tracking algorithm for guitarist based on temporal grouping and pattern analysis

    AU - Wang, Zhao

    AU - Ohya, Jun

    PY - 2017

    Y1 - 2017

    N2 - This paper proposes a temporal Grouping and pattern analysis-based algorithm that could track the fingertips of guitarists during their guitar playing towards the actualization of the automatic guitar fingering recognition system. First a machine learning-based Bayesian Pixel Classifier is used to segment the hand area on the test data. Then, the probability map of fingertip is generated on the segmentation results by counting the voting numbers of the Template Matching and Reversed Hough Transform. Furthermore, a temporal Grouping algorithm, which is a geometry analysis for consecutive frames, is applied to removal noise and group the same fingertips (index finger, middle finger, ring finger, little finger). Then, a data association algorithm is utilized to associate 4 tracked fingers (index finger, middle finger, ring finger, little finger) with their correspondent tracked results frame by frame. Finally, particles are distributed only between the associated fingertip candidates to track the fingertips of guitarist effectively. The experimental result demonstrates that this fingertip tracking algorithm is robust enough for tracking fingertips (1) without any constrains such us color marker; (2) under the complex contexts, such us complicated background, different illumination conditions, (3) with the high tracking accuracy (mean error 3.36 pixels for four fingertips).

    AB - This paper proposes a temporal Grouping and pattern analysis-based algorithm that could track the fingertips of guitarists during their guitar playing towards the actualization of the automatic guitar fingering recognition system. First a machine learning-based Bayesian Pixel Classifier is used to segment the hand area on the test data. Then, the probability map of fingertip is generated on the segmentation results by counting the voting numbers of the Template Matching and Reversed Hough Transform. Furthermore, a temporal Grouping algorithm, which is a geometry analysis for consecutive frames, is applied to removal noise and group the same fingertips (index finger, middle finger, ring finger, little finger). Then, a data association algorithm is utilized to associate 4 tracked fingers (index finger, middle finger, ring finger, little finger) with their correspondent tracked results frame by frame. Finally, particles are distributed only between the associated fingertip candidates to track the fingertips of guitarist effectively. The experimental result demonstrates that this fingertip tracking algorithm is robust enough for tracking fingertips (1) without any constrains such us color marker; (2) under the complex contexts, such us complicated background, different illumination conditions, (3) with the high tracking accuracy (mean error 3.36 pixels for four fingertips).

    UR - http://www.scopus.com/inward/record.url?scp=85016126960&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85016126960&partnerID=8YFLogxK

    U2 - 10.1007/978-3-319-54526-4_16

    DO - 10.1007/978-3-319-54526-4_16

    M3 - Conference contribution

    AN - SCOPUS:85016126960

    SN - 9783319545257

    VL - 10118 LNCS

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

    SP - 212

    EP - 226

    BT - Computer Vision - ACCV 2016 Workshops, ACCV 2016 International Workshops, Revised Selected Papers

    PB - Springer Verlag

    ER -