OpenTUG.zip

The material published here comes from the currently submitted manuscript : "Walking towards the future – Exploring OpenTUG's validity in automatic walking activity analyses and the relationship with cognition in vestibular patients." by Emilie Lacroix, Marius Grandjean, Margaux Huyberechts, Lucie Steenbergen, Seyed Abolfazl Ghasemzadeh, Christophe De Vleeschouwer, Naïma Deggouj, Martin Gareth Edwards.

OpenTUG builds on OpenPifPaf (Kreiss et al., 2019), a bottom-up method for multi-person 2D human pose estimation, predicting timing and the number of steps of the person from an accurate skeleton estimation of the human body. See paper for more information.

Kreiss, S., Bertoni, L., & Alahi, A. (2019). PifPaf: Composite Fields for Human Pose Estimation. http://arxiv.org/abs/1903.06593