A dataset of high-resolution plantar pressures for gait analysis across varying footwear and walking speeds

TL;DR

The UNB StepUP-P150 dataset is introduced: a footStep database for gait analysis and recognition using Underfoot Pressure, including data from 150 individuals, establishing a new benchmark for plantar pressure-based gait analysis and recognition.

Abstract

Gait refers to the patterns of limb movement generated during walking, which are unique to each individual due to both physical and behavioral traits. Walking patterns have been widely studied in biometrics, biomechanics, sports, and rehabilitation. While traditional methods rely on video and motion capture, advances in plantar pressure sensing technology now offer deeper insights into gait. However, underfoot pressures during walking remain underexplored due to the lack of large, publicly accessible datasets. To address this, we introduce the UNB StepUP-P150 dataset: a footStep database for gait analysis and recognition using Underfoot Pressure, including data from 150 individuals. This dataset comprises high-resolution plantar pressure data (4 sensors per cm-squared) collected using a 1.2m by 3.6m pressure-sensing walkway. It contains over 200,000 footsteps from participants walking with various speeds (preferred, slow-to-stop, fast, and slow) and footwear conditions (barefoot, standard shoes, and two personal shoes), supporting advancements in biometric gait recognition and presenting new research opportunities in biomechanics and deep learning. UNB StepUP-P150 establishes a new benchmark for plantar pressure-based gait analysis and recognition.

Figures (16)

• Figure 1: Distributions of (a) participants' ages by sex and (b) participants' chosen standard shoe sizes by sex, in UK sizes. Note: Female distributions are shown in orange, and male distributions are shown in dark blue. A Wilcoxon rank-sum test did not indicate any significant difference in the distribution of ages for the female and male subgroups ($p = 0.55$). The shoe sizes for the male subgroup were significantly larger than the female subgroup ($p < 0.0001$ using a two sample $t$-test).
• Figure 2: Distributions of participants' physical characteristics by sex and ethnicity/race: (a) height (cm), weight (kg), and foot size (marker size is proportional to measured foot length in cm), and (b) measured foot length (cm) and width (cm). Note: Orange markers are used for female participants and dark blue for male participants. Some jitter was added for (b) to improve visibility.
• Figure 3: Overview of the instrumentation configuration. (a) A diagram of the laboratory setup; participants walked back and forth across a grid of twelve sensing tiles encircled by a non-instrumented platform to allow for turning. Seven RGB video cameras were used to capture the participants from different viewing angles. (b) A video frame from Camera 7 during a walking trial with corresponding pressure measurements.
• Figure 4: Overview of the experimental protocol. After a 30 minute preparation period for onboarding and familiarizing the participant with the study, three 30-second standing trials (S1, S2, and S3) and four 90-second walking trials (W1, W2, W3, and W4) were recorded for each of the four footwear conditions (BF, ST, P1, and P2). The participants were allowed to take breaks throughout the study as needed, with at least two minutes taken to sit down and change shoes between footwear conditions. The order of the footwear conditions and walking speeds were randomized for each participant.
• Figure 5: Average walking speeds, categorized by sex, as computed from pressure measurements during each walking task of the experimental procedure, and averaged over the four different footwear conditions. Note: The slow-to-stop, slow, and fast walking speeds were all found to be significantly different than the participants' preferred walking speeds ($p < 0.05$ using paired $t$-tests). There were no statistically significant differences between walking speeds for female and male subgroups ($p > 0.05$ for all walking tasks using two-sample $t$-tests)