Full-Body 3D Human Gait Dataset walking on flat ground

This dataset contains full-body 3D gait data collected from 26 healthy participants (10 males, 16 females) with an average age of 28.19 ± 7.77 years. Data was captured using the Xsens Awinda MTw inertial measurement system, comprising 17 wireless sensors operating at a 60Hz sampling frequency.

Key Features:

Full-body motion data using MVN Analyze software's full-body model

Anthropometric measurements: height (170.5 ± 8.61 cm), foot length (26.47 ± 1.88 cm), shoulder width (39.32 ± 7.79 cm), and wrist span (131.36 ± 8.85 cm)

Four distinct walking paths: Mixed (straight and curved), Circle (3m diameter), Turn (180-degree turns), and Zigzag

Total of 1,024,295 frames (17,071.58 seconds) of gait recordings

Average of 3,568.97 ± 1,204.26 frames per recording (59.48 ± 20.07 seconds)

The dataset includes various walking patterns designed to capture a wide range of gait characteristics, including straight walks, gentle curves, sharp turns, and zigzag movements. Participants were allowed some freedom in executing turns, particularly in the Zigzag and Mixed paths, to introduce natural variations in gait patterns.

This comprehensive dataset is suitable for gait analysis, biomechanics research, and the development of motion synthesis algorithms, particularly those focused on normal walking patterns on a fixed surface with various turning scenarios.

Dataset Structure:

'participants.xlsx': An Excel file containing participant codes and their anthropometric data.

'data' folder: Contains subdirectories named with participant codes.

Each participant subdirectory contains CSV files of different gait recordings for that participant.

This dataset was collected as part of the study:

Carneros-Prado, D., Dobrescu, C. C., Cabañero, L., Villa, L., Altamirano-Flores, Y. V., Lopez-Nava, I. H., … & Hervás, R. (2024). Synthetic 3D full-body skeletal motion from 2D paths using RNN with LSTM cells and linear networks. Computers in Biology and Medicine, 180, 108943.