SAIL-TUG

Overview

Timed-Up and Go video for SAIL(SAIL-TUG) is a new medical gait data set, focus on abnormal gait detection of person. Following are the detailed descriptions.

• Volume
  The dataset contains 404 TUG videos. By downsampling, 38,098 images is annotated with 2D human poses with 16 pixel-wise key points.
• Diversity
  The 404 TUG videos are recored in several different places including hospitals, communities and nursing homes. And the gait of the subjects also varied widely.

Downloads

Please contact the author for the original data.

Example

• Video example
  

• Image example

  

• Annotation points

  Annotation format:

  {
      "info": {'year': 2019,
               'version': 0.1,
               'description': None,
               'contributor': 'Yanhong Wang',
               'url': None,
               'date_created': '2019-05-29'},
      "licenses": {'url': 'http://creativecommons.org/licenses/by-nc-sa/2.0/',
    'name': 'Attribution-NonCommercial-ShareAlike License'},
      "imgname": [],  #imgname.shape:(30077, 16)
      "bndbox": [],  #bndbox.shape:(30077, 1, 4)
      "part": []   #parts.shape:(30077, 17, 2)
  }
  

  Key points visualization example:

  

  Key points list:

  No.	0	1	2	3	4	5	6
  Key points	Nose	LEye	REye	LEar	REar	LShoulder	RShoulder
  No.	7	8	9	10	11	12	13
  Key points	LElbow	RElbow	LWrist	RWrist	LHip	RHip	LKnee
  No.	14	15	16
  Key points	RKnee	LAnkle	RAnkle

  Note: $LEye$ means left eye and the same for other joints.

• Gait parameters annotation:

  These are video-level annotations which means every each video corresponds a set of parameters. And there are 404 sets in this dataset.

  

  Two sets of annotation example:

  

  Note: The parameters in Table III are not completely manually tagged because of resource factors, which has little inference to the experiments.

Citation

{
@article{SAIL...,
author       = {Yanhong Wang},
title        = {SAIL: A Smart AI Algorithm Framework For Automatic Gait Assessment Based On TUG Test},
month        = ,
year         = ,
doi          = {},
version      = {1.0},
url          = {}
}

Please also email([email protected]) us with the title of your paper or work on the dataset.