search

erikbohnsack / pmbm

Python implementation of Poisson Multi-Bernoulli Mixture Filter for Multi-Object Tracking.
64 stars 23 forks source link

Is it a real pmbm algorithms? #9

Open SmallMunich opened 2 years ago

SmallMunich commented 2 years ago

Hi, @erikbohnsack . thanks for supply this code. However, I confused this algorithm is pmbm ? because I found the code about possion birth is not valid. I think this code is mbm algorithm, is it rightly?

erikbohnsack commented 2 years ago

Hello @SmallMunich!

This starts to become quite some time ago. I remember that we choose to go away from a Gaussian-composed density birth model to a simple uniform distribution birth model due to simplicity and shortage of time, however I think one could still argue that it was a Poisson model. Could you specify how you think it is invalid?

Best, Erik

AlexanderMiles commented 2 years ago

Hey @erikbohnsack, first of all thanks for providing this code. I am currently learning RFS methods for tracking and this has been very helpful to see the implementation of some of the theory.

I noticed same as SmallMunich that the uniform approach essentially skips the Poisson creation, this means you never get to a situation where you have Poisson distributions being tracked? Therefore you will always run the else statement below the comment " # If no poisson components, then just go with the uniform measurement state.".

I also notice that RFS methods talk about not requiring data association however this method does this via Murty's algorithm. Would I be right in saying that without needing to keep track of the target ids you could essentially skip this step and just keep essentially active tracks with no past history.

rakesh2024 commented 7 months ago

How can i execute this on my computer . please some one explain me