From Image to Stability: Learning Dynamics from Human Pose (acceptance rate:26%)

We propose and validate two end-to-end deep learning architectures to learn foot pressure distribution maps (dynamics) from 2D or 3D human pose (kinematics). The networks are trained using 1.36 million synchronized pose+pressure data pairs from 10 subjects performing multiple takes of a 5-min long choreographed Taiji sequence. Using leave-one-subject-out cross validation, we demonstrate reliable and repeatable foot pressure prediction, setting the first baseline for solving a non-obvious pose to pressure cross-modality mapping problem in computer vision. Furthermore, we compute and quantitatively validate Center of Pressure (CoP) and Base of Support (BoS), two key components for stability analysis, from the predicted foot pressure distributions.

This file is the accepted manuscript. Use is subject to Springer's Accepted Manuscript Terms of Use.



Work Title From Image to Stability: Learning Dynamics from Human Pose (acceptance rate:26%)
Open Access
  1. Jesse Scott
  2. Bharadwaj Ravichandran
  3. Christopher Funk
  4. Robert T. Collins
  5. Yanxi Liu
License In Copyright (Rights Reserved)
Work Type Article
  1. Springer International Publishing
Publication Date 2020
Publisher Identifier (DOI)
  1. 10.1007/978-3-030-58592-1_32
  1. Computer Vision – ECCV 2020
  2. Lecture Notes in Computer Science
Deposited September 22, 2022




This resource is currently not in any collection.

Work History

Version 1

  • Created
  • Added ECCV2020_SCOTT_etal-1.pdf
  • Added Creator Jesse Scott
  • Added Creator Bharadwaj Ravichandran
  • Added Creator Christopher Funk
  • Added Creator Robert T. Collins
  • Added Creator Yanxi Liu
  • Published