Home > Research > Publications & Outputs > Learning Human Pose Models from Synthesized Dat...

Electronic data

  • Learning_Human_Pose_Models_from_Synthesized_Data_for_Robust_RGB_D_Action_Recognition__IJCV_Revision_2_

    Rights statement: The final publication is available at Springer via http://dx.doi.org/10.1007/s11263-019-01192-2

    Accepted author manuscript, 7.04 MB, PDF document

    Embargo ends: 6/08/20

    Available under license: CC BY-NC: Creative Commons Attribution-NonCommercial 4.0 International License

Links

Text available via DOI:

View graph of relations

Learning Human Pose Models from Synthesized Data for Robust RGB-D Action Recognition

Research output: Contribution to journalJournal article

E-pub ahead of print
Close
<mark>Journal publication date</mark>6/08/2019
<mark>Journal</mark>International Journal of Computer Vision
Number of pages20
Publication statusE-pub ahead of print
Early online date6/08/19
Original languageEnglish

Abstract

We propose Human Pose Models that represent RGB and depth images of human poses independent of clothing textures, backgrounds, lighting conditions, body shapes and camera viewpoints. Learning such universal models requires training images where all factors are varied for every human pose. Capturing such data is prohibitively expensive. Therefore, we develop a framework for synthesizing the training data. First, we learn representative human poses from a large corpus of real motion captured human skeleton data. Next, we fit synthetic 3D humans with different body shapes to each pose and render each from 180 camera viewpoints while randomly varying the clothing textures, background and lighting. Generative Adversarial Networks are employed to minimize the gap between synthetic and real image distributions. CNN models are then learned that transfer human poses to a shared high-level invariant space. The learned CNN models are then used as invariant feature extractors from real RGB and depth frames of human action videos and the temporal variations are modelled by Fourier Temporal Pyramid. Finally, linear SVM is used for classification. Experiments on three benchmark human action datasets show that our algorithm outperforms existing methods by significant margins for RGB only and RGB-D action recognition.

Bibliographic note

The final publication is available at Springer via http://dx.doi.org/10.1007/s11263-019-01192-2