GAIT-IT dataset
About:
The GAIT-IT dataset contains the simulation of pathological gait, to be used for testing gait analysis and classification systems.
The dataset was recorded in the professional studio of FCT|FCCN (Fundação para a Ciência e a Tecnologia), on two different days, with a chroma key green background and under artificial lighting, allowing the extraction of high-quality binary silhouettes. The side-view camera was placed at approximately 3 meters from the target.
The GAIT-IT dataset contains simulations of 5 different types of gait: (i) the scissor gait, commonly associated with diplegia affects both sides of the body; (ii) the spastic gait, commonly associated with hemiplegia affects only one side of the body; (iii) the steppage gait, commonly associated with neuropathy leads to foot drop; and (iv) the propulsive gait, commonly associated with Parkinson’s diseases is characterized by a stooped posture, with both arms held close to the chest and the lower limbs flexed and rigid.
For each type, except normal, two levels of severity are captured. The subjects provide 4 gait sequences per severity level. This corresponds to a subject walking twice from left to right and from right to left, from the side view. The sequences are captured on two different days where 21 volunteers (19 males and 2 females) in the age range of 20 to 56 years. 2 participants repeated the experience on the second day, leading to a total of 828 gait sequences. Before capturing the sequences subjects were instructed on how to simulate the various gait types and severity levels.
Details:
The database is organized in the following hierarchical folder structure:
· The main directory, “GAIT-IT”, contains 5 folders corresponding to each of the 5 gait types featured in the dataset named: diplegic (scissor), hemiplegic (spastic), neuropathic (steppage), normal and Parkinson (propulsive).
· Each of these 5 gait type folders has 23 folders corresponding to all subject performances (s1, …, s23).
· For every subject, there are 4 folders corresponding to each input type featured in the dataset, namely, Binary Silhouettes, Binary Skeletons, GEIs and SEIs. These are named: silhouettes, skeletons, GEIs, SEIs.
In the case of pathological gait sequences, each of these subfolders is named subipatjlvlk-w_direction, where:
· i is the subject number
· j is the pathology number corresponding to one of the gait related pathologies (1: scissor /diplegic, 2: spastic/hemiplegic, 3: steppage/neuropathic, 4: propulsive/parkinsonian).
· k can be 1 or 2, identifying the severity level, where 2 represents the most severe.
· w can also be 1 or 2 identifying one of the two repetitions for each sequence.
· direction can be front or back and represents the walking direction.
In the case of normal gait, the subfolders are named as subinormal-w_direction.
License agreement:
· To use the database please fill in
the license agreement and send a
scanned copy of the signed form by e-mail.
Download:
·
The database
can be downloaded from the links below.
·
A password for decrypting the compressed
Zip files will be provided after receiving the duly signed license agreement
(see above).
Download:
· Please contact the authors by e-mail.
How to reference the database:
· P. Albuquerque, J. Machado, T. Verlekar, P. L. Correia; L. D. Soares, “Remote Gait Type Classification System using Markerless 2D Video”, MDPI Diagnostics, 2021, https://doi.org/10.3390/diagnostics11101824
Contacts:
Feedback is
welcome. Please provide your comments and suggestions by e-mail.