A Comprehensive Analysis of Deep Regression

Item request has been placed! ×
Item request cannot be made. ×
loading   Processing Request
  • Additional Information
    • Publication Information:
      USA: IEEE
    • Publication Date:
      2020
    • Abstract:
      Deep learning revolutionized data science, and recently its popularity has grown exponentially, as did the amount of papers employing deep networks. Vision tasks, such as human pose estimation, did not escape from this trend. There is a large number of deep models, where small changes in the network architecture, or in the data pre-processing, together with the stochastic nature of the optimization procedures, produce notably different results, making extremely difficult to sift methods that significantly outperform others. This situation motivates the current study, in which we perform a systematic evaluation and statistical analysis of vanilla deep regression, i.e., convolutional neural networks with a linear regression top layer. This is the first comprehensive analysis of deep regression techniques. We perform experiments on four vision problems, and report confidence intervals for the median performance as well as the statistical significance of the results, if any. Surprisingly, the variability due to different data pre-processing procedures generally eclipses the variability due to modifications in the network architecture. Our results reinforce the hypothesis according to which, in general, a general-purpose network (e.g., VGG-16 or ResNet-50) adequately tuned can yield results close to the state-of-the-art without having to resort to more complex and ad-hoc regression models.
    • Author Affiliations:
      PERCEPTION Team, Inria, Université Grenoble Alpes, Montbonnot, France
    • ISSN:
      0162-8828
      2160-9292
      1939-3539
    • Accession Number:
      10.1109/TPAMI.2019.2910523
    • Rights:
      Copyright 1979-2012, IEEE
    • AMSID:
      8686063
    • Date of Current Version:
      2020
    • Document Subtype:
      IEEE Transaction
    • Sponsored by:
      IEEE Computer Society
    • Accession Number:
      edseee.8686063
  • Citations
    • ABNT:
      LATHUILIERE, S. et al. A Comprehensive Analysis of Deep Regression. IEEE Transactions on Pattern Analysis and Machine Intelligence, Pattern Analysis and Machine Intelligence, IEEE Transactions on, IEEE Trans. Pattern Anal. Mach. Intell, [s. l.], v. 42, n. 9, p. 2065–2081, 2020. DOI 10.1109/TPAMI.2019.2910523. Disponível em: http://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edseee&AN=edseee.8686063. Acesso em: 1 dez. 2020.
    • AMA:
      Lathuiliere S, Mesejo P, Alameda-Pineda X, Horaud R. A Comprehensive Analysis of Deep Regression. IEEE Transactions on Pattern Analysis and Machine Intelligence, Pattern Analysis and Machine Intelligence, IEEE Transactions on, IEEE Trans Pattern Anal Mach Intell. 2020;42(9):2065-2081. doi:10.1109/TPAMI.2019.2910523
    • APA:
      Lathuiliere, S., Mesejo, P., Alameda-Pineda, X., & Horaud, R. (2020). A Comprehensive Analysis of Deep Regression. IEEE Transactions on Pattern Analysis and Machine Intelligence, Pattern Analysis and Machine Intelligence, IEEE Transactions on, IEEE Trans. Pattern Anal. Mach. Intell, 42(9), 2065–2081. https://doi.org/10.1109/TPAMI.2019.2910523
    • Chicago/Turabian: Author-Date:
      Lathuiliere, S., P. Mesejo, X. Alameda-Pineda, and R. Horaud. 2020. “A Comprehensive Analysis of Deep Regression.” IEEE Transactions on Pattern Analysis and Machine Intelligence, Pattern Analysis and Machine Intelligence, IEEE Transactions on, IEEE Trans. Pattern Anal. Mach. Intell 42 (9): 2065–81. doi:10.1109/TPAMI.2019.2910523.
    • Harvard:
      Lathuiliere, S. et al. (2020) ‘A Comprehensive Analysis of Deep Regression’, IEEE Transactions on Pattern Analysis and Machine Intelligence, Pattern Analysis and Machine Intelligence, IEEE Transactions on, IEEE Trans. Pattern Anal. Mach. Intell, 42(9), pp. 2065–2081. doi: 10.1109/TPAMI.2019.2910523.
    • Harvard: Australian:
      Lathuiliere, S, Mesejo, P, Alameda-Pineda, X & Horaud, R 2020, ‘A Comprehensive Analysis of Deep Regression’, IEEE Transactions on Pattern Analysis and Machine Intelligence, Pattern Analysis and Machine Intelligence, IEEE Transactions on, IEEE Trans. Pattern Anal. Mach. Intell, vol. 42, no. 9, pp. 2065–2081, viewed 1 December 2020, .
    • MLA:
      Lathuiliere, S., et al. “A Comprehensive Analysis of Deep Regression.” IEEE Transactions on Pattern Analysis and Machine Intelligence, Pattern Analysis and Machine Intelligence, IEEE Transactions on, IEEE Trans. Pattern Anal. Mach. Intell, vol. 42, no. 9, Sept. 2020, pp. 2065–2081. EBSCOhost, doi:10.1109/TPAMI.2019.2910523.
    • Chicago/Turabian: Humanities:
      Lathuiliere, S., P. Mesejo, X. Alameda-Pineda, and R. Horaud. “A Comprehensive Analysis of Deep Regression.” IEEE Transactions on Pattern Analysis and Machine Intelligence, Pattern Analysis and Machine Intelligence, IEEE Transactions on, IEEE Trans. Pattern Anal. Mach. Intell 42, no. 9 (September 1, 2020): 2065–81. doi:10.1109/TPAMI.2019.2910523.
    • Vancouver/ICMJE:
      Lathuiliere S, Mesejo P, Alameda-Pineda X, Horaud R. A Comprehensive Analysis of Deep Regression. IEEE Transactions on Pattern Analysis and Machine Intelligence, Pattern Analysis and Machine Intelligence, IEEE Transactions on, IEEE Trans Pattern Anal Mach Intell [Internet]. 2020 Sep 1 [cited 2020 Dec 1];42(9):2065–81. Available from: http://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edseee&AN=edseee.8686063