Skip to content

A particle-filter algorithm to perform visual tracking with an event camera

License

Notifications You must be signed in to change notification settings

Aniket-Gujarathi/gaze-tracking

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Event-driven Gaze Tracking with a Particle Filter

The algorithm performs tracking of eye-gaze target in-the-wild given the event-stream from an event-camera.

CPU

A C++ YARP module that runs the algorithm on a standard CPU.

How to build

  1. install YARP
  2. install event-driven
  3. make sure both YARP and event-driven can be found on the environment path / project path
  4. clone this repository (e.g. into ~/projects/particle-filter-tracking)
  5. cd ~/projects/particle-filter-tracking && mkdir build && cd build
  6. cmake ..
  7. make
  8. ensure the bin is found in the environment path

How to run

  1. yarpserver --write
  2. yarpmanager
  3. open the provided yarpmanager application (particle-filter-tracking/cpu/gaze_tracking.xml)
  4. run all
  5. open a dataset with the yarpdataplayer
  6. connect all
  7. play the dataset on the yarpdataplayer

Note: you can substitute the yarpdataplayer with a live feed of an event-camera if it is yarp-compatible. Please contact the authors for more information if you want to make your camera yarp compatible.

Implementation

  1. The --src folder contains two implementations for the circle-target based eye-ball tracking and the parabola-based eyebrow tracking (for efficient search space).
  2. Compared to related methods, the tracking algorithm does not assume a fixed camera or head of a person.

Results

Parabola Tracking
video01.mp4
Circle Tracking
video_left.mp4

About

A particle-filter algorithm to perform visual tracking with an event camera

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 71.2%
  • C++ 23.6%
  • MATLAB 4.8%
  • CMake 0.4%