Skip to content
master
Switch branches/tags
Go to file
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 

Classroom Person Tracker

Final Project for Advances in Computer Vision (at MIT)

See paper_submission.pdf for an outline of the project's background, development process, and results.

In this project, I devise a set of algorithms that utilize motion-based tracking techniques to create compelling visualizations such as a dynamic heat maps of people's physical location and movement. The algorithms are loosely based on the Kalman filter and Lucas-Kanade methods. The performance of the algorithms is demonstrated using our lab's recordings of students and teachers from ceiling-mounted webcams in Montessori preschool classrooms. The objective is to use the data to report meaningful information about the students' habits in the schools by creating visualizations that would help teachers better understand how the physical space is utilized.

Code

  • lucasKanadeCentroids.m is the algorithm that uses Lucas Kanade method to find clusters of optical flow vectors, and places a marker on the centroid of the cluster vectors.
  • lucasKanadeBounding.m is the algorithm that uses Lucas Kanade method to create bounding boxes around every optical flow vector, and merges overlapping bounding boxes to construct a final bounding box for each tracked subject in the frame.
  • multiObjectTracking.m is a slightly modified version of this Matlab example: http://www.mathworks.com/help/vision/examples/motion-based-multiple-object-tracking.html.

The .mat files contain results obtained using these scripts, which we utilized when constructing the graphs located in the "figures" directory.

Two demo videos are attached: mutliple_people and one_person, which can be used with any of the scripts attached to recreate the visual representations discussed in the paper.

About

Motion tracking of people from classroom webcam footage

Resources

Releases

No releases published

Packages

No packages published

Languages