TracTrac: PTV software.

TracTrac PTV software _______________


TracTrac is a Particle Tracking Velocimetry (PTV) software which is fast (more than 10000 points tracked per second) and accurate (up to 0.01 pixel resolution), forming thus a good concurrent to the state-of-the art PIV algorithms. It allows to track anything that moves: birds, ants, grains, water flows... It runs on Python (v2.7 with OpenCV2) or Matlab (>2012a with Image Processing toolbox). Give it a try !

GitHub repository_________________________________

> Get the last sources for TracTrac on GitHub!

SourceForge Release ______________________________

> Download : Last release / Older Versions
Report a bug

Reference _________________________________

> Please refer to the following work :
Heyman J., TracTrac: A fast multi-object tracking algorithm for motion estimation, Computers & Geosciences, Volume 128, 2019, Pages 11-18,doi: 10.1016/j.cageo.2019.03.007 , PDF.

User Guide _________________________________

- Matlab users, run gui with
>> tractrac

- Python users, run test cases with
>> python tractrac.py -p 1
or
>> python tractrac.py -f '../Sample_videos/videotest.avi' -p 1
or
>> python tractrac.py -f '../Sample_videos/PIVChallenge/*.tif' -p 2
or
>> python tractrac.py -f '../Sample_videos/RiverDrone/*.tif' -p 2

- Run live-treatment by webcam acquisition by
>> python tractrac.py -f '0' -p 1 -cmax 50

Details of parameters and commands are given in the publication Heyman 2019.

Examples ________________________________________

Python example 1:
Visualisation
Matlab Example 1:
Granular Flow from hopper
Matlab Example 2:
Bird Flock
Matlab Example 3:
Synthetic PIV flow

MATLAB GUI for TracTrac ________________________

TracTrac comes out with a sample video of grains in motion. Simply load it, and click Start to begin computations. TracTrac detects peaks in the video frame and predicts their trajectories through time.
TracTrac is able to simultaneously track several thousand of peaks, this in very short time (about 1 second per frame for 10000 peaks).
Such efficiency is reached by splitting the association process into several ones, over sub-windows.
TracTrac is accurate, even in regions of high velocity gradients. A sub-pixel algorithm coupled to a motion model allows it to track particles of velocities ranging from 0 to several pixels per frames.
TracTrac is flexible and user-friendly. There are several visualization options, that allows you to adjust in real-time the tracking parameters to reach the best results.
Parameters can be conveniently saved for later use.
Tractrac integrates manual cropping of videos, associated to median background subtraction, in order to focus better on image regions of direct interest.
Rotation is also possible in version 1.2.
Post-processing is also available inside the TracTrac program. Several plotting options are available: average velocities, fluctuations, peak density and peak trajectories, for instance.
Figures may be saved with a single click ! Convinced ?!