This is a python port and visualization of the UI-PRMD dataset/paper This Port Is created for a project i am currently devloping which is in python hence a python port was necessary
UI-PRMD is a data set of movements related to common exercises performed by patients in physical therapy and rehabilitation programs.
- the paper contains code visualization of the movements written MATLAB which I found cumbersome use I adapted the the code and implemented it in python
- for easy understanding of how the dataset is represented use this port
- use it to convert the 2 file dataset into a single 2D matrix of cartesian cordinates
the following libraries need to be installed
- python 3.6
- numpy
- matplotlib
- pandas
- celluloid
copy the following while creating the virtual env
pip install matplotlib
pip install pandas
pip install numpy
pip install celluloid
(rendered in matplotlib)
within the segmented movements(download from website link) folder are many movements their positions and corresponding angles m01_s01_e02_angles.txt
corresponds to movement 1 subject 1 episode 2's angle inforamtion if read as an array it would be something like
| joint1 X angle | joint1 Y angle | joint1 Z angle | joint2 X angle | . | . | |
|---|---|---|---|---|---|---|
| frame1 | data | data | data | data | data | data |
| frame2 | data | data | data | data | data | data |
| . | data | data | data | data | data | data |
| . | data | data | data | data | data | data |
| . | data | data | data | data | data | data |
[frames x 66] (22 joints x 3 axis = 66)
similarly it is same for the positions file
as the angles are YXZ triplet of Euler angles we combine it with the positions and after all operations the resulting and returned numpy array shape is [22 X 3 X frames] this is plotted as a skeleton in matplotlib
the 3 columns represent the x,y,z values x-left, right y-up,down, z- in and out of screen
each row is a joint
depth is the frames the kinect used to capture data is 30FPS
- joint list
Waist,Spine,Chest,Neck,Head,Head tip,Left Collar,Left Upper arm,Left for arm,Left hand,Right collar,Right upper arm,Right forearm,Right hand,Left upper leg,Left lower leg,Left foot,Left leg toes Right upper leg,Right lower leg,Right foot
*movemet list
(a) Deep squat (m01); (b) Hurdle step (m02); (c) Inline lunge (m03); (d) Side lunge (m04); (e) Sit to stand (m05); (f) Standing active straight leg raise (m06); (g) Standing shoulder abduction (m07); (h) Standing shoulder extension (m08); (i) Standing shoulder internal-external rotation (m09); (j) Standing shoulder scaption (m10)