Optical flow & face recognition and tracking server

Optical flow & face recognition and tracking server

Closed - This job posting has been filled and work has been completed.

Job Description

I'm working on interactive installation where interactivity will be based only on inputs from web cam.

I would like to create a simple application probably based on OpenCV that will generate based on live web camera, text output about position of faces and if possible somehow about strongest optical flow. Text info could be in JSON format and I would like to serve probably via Node.js (I can do this part).

Example:
faces: { head1: { x:30, y:50, w:100, h:180}, head2: {......

flow: { id1: { x:40, y:100, angle: 45, speed: 6}... ?? - I'm open to any ideas how to track optical flow, I just would like to keep it simple and short output


The data from motion I will use in my web application for Augmented Reality.


As video input I can offer webcam, dual web cam or Kinect (or combination of all)

If you will be interested in development I can send you a video sample of location with people there and some movement.

I have found http://opentsps.com/ as close example application however I find it a bit difficult to adjust it for correct setup.

I found very interesting open source for fast face/head tracking: http://ros.org/wiki/pi_face_tracker but I don't have skills to use it.


Requirements:

multiple face tracking from live video
must run on mac os x
standalone app or terminal command line
output text info in JSON (file or websocket in Node.js)

If you have any question please contact me.

Best Regards,

Albert

---
Skills: mac-os-x, json, video