Improving Speed x1.5 - Githubissues

abewley / sort

Simple, online, and realtime tracking of multiple objects in a video sequence.

GNU General Public License v3.0

3.82k stars 1.07k forks source link

Open laclouis5 opened 4 years ago

laclouis5 commented 4 years ago

You can even more improve execution speed by passing compute_log_likelihood=False flag when initializing KalmanFilter line 85.

This improve speed x1.5.

abewley commented 4 years ago

Great find! I'll add this to my dev branch which I should hope to merge soon.

tylernewnoise commented 4 years ago

I think this will not work with filterpy 1.4.5 and was fixed/changed here.