The University of Adelaide
33 files

ForceID Dataset

posted on 2021-09-29, 04:37 authored by Kayne DuncansonKayne Duncanson, Simon ThwaitesSimon Thwaites, David Booth, Ehsan AbbasnejadEhsan Abbasnejad, William Robertson, Dominic ThewlisDominic Thewlis
This dataset was acquired for research on the use of gait as a (soft) biometric for person re-identification/recognition; however, it may be used to answer a variety of research questions.

Experimental protocol:
The dataset was generated through a repeated measures experiment (approved by the Human Research Ethics Committee - approval No. H-2018-009) conducted at The University of Adelaide gait analysis laboratory. Participants were recruited from the general population if they were in good health and had no known neurological disorder that could affect their gait (e.g. stroke, cerebral palsy, Parkinson's disease, etc.). The experiment was conducted across two sessions separated by a minimum of 3 and a maximum of 14 days. Participants wore their own personal clothing and footwear, though footwear had to be consistent between sessions. Age, sex, mass, height, and footwear type were recorded at the start of each session (footwear was photographed for future reference). Body mass was measured using standard digital scales and height was measured using a stadiometer. Participants completed five trials of walking along the length of the laboratory (10m) at preferred walking speed. Embedded into the floor surface at the center of the laboratory, two OPT400600-HP force platforms (Advanced Mechanical Technology Inc., USA) measured ground reaction forces (GRFs) and moments during left and right footsteps (i.e. stance phases). These measures, along with the calculated center of pressure (COP) coordinates, were acquired through Vicon Nexus at 2000Hz. Two Vicon Vue video cameras (Vicon, Oxford, UK) were time synchronized with the force platforms and captured motion from front and side views. The video footage was acquired at 50Hz with 1920x1080 pixel resolution so that foot contact regions and stance sides could be clearly identified. Trials were included in the dataset if each foot contacted within the sensing area of each force platform, as identified from the video footage.

Dataset characteristics:
In total, there were 1087 trials from 111 participants with 8-10 trials per participant. From each trial there were eight 1D time series signals per stance side representing individual components of the GRF (Fx, Fy, Fz) (N), moment (Mx, My, Mz) (, and COP (Cx, Cy) (mm). Two versions of the dataset are presented here: the raw version as exported from Vicon Nexus and the pre-processed version based on the method outlined in [publication TBC]. The dataset was organised into separate files according to measure, component, stance side, and version (respectively). In each data file are participant ID labels (column one), session numbers (column two), and respective measurements (column three onward). Finally, there is a metadata file that contains the age (years), sex (Male M / Female F), body mass (kg), height (m), and footwear category for each identity. Body mass and height are reported as the average measure across sessions one and two.


Improving the functional outcomes of lower limb orthopaedic surgery

National Health and Medical Research Council

Find out more...

Australian Government Research Training Program Stipend (RTPS)

Defence Science and Technology Group