Project SitApp

Maroun Sassine
Asher Yanai

Project abstract

Project Sit is a project that will help every user to sit properly while using his computer. The application will notify the user when his sitting posture is wrong so he can correct it and keep his back and neck healthy. The application uses Intel RealSense camera to gather depth information and some features of its SDK.

The application starts by first recording and gathering the information of the best sitting posture the user can be in. After gathering all information, the camera will get the live user sitting posture and compare it with the information saved from the recorded scene.

Introduction:

To estimate the shoulders’ location, we used the human face golden ratio. The Intel SDK gave us 3 points ( right eye, left eye and chin).

Neufert_Big_Man

*The body golden ratio.

How does it work ?

  • First of all, the app will search for a face using Intel RealSense SDK.
  • If a face was detected, get a segmented image using the SDK.
  • Now we can look for shoulders!
  • Transform the image to a binary image and find the edge of this image.
  • Now using the golden ratio and 3 points (left eye, right eye and chin), we can find intersection points of the circle with the shoulders.
  • Now we can search for the shoulders points and get the depth.

Algorithm picture sequence

binary

*The binary image

faceShoulderRatio

*Golden ratio used to find intersection points between circle and shoulders

After finding the intersection points, now we search in the region and find the farthest point from the linear line that intersects with the intersection point we found earlier and the beginning of the edge.

shoulderDetection

The shoulder plot “edge function”.

shoulderedge

Shoulder detection result

Video

Source code

Soure code

Run ProjectSit.exe found in bin -> debug. For more help see the readme file.

References

Accessibility
Close Close