# Multiple object tracking package

A simple multiple object tracking demonstration in the Birch probabilistic programming language. This demonstrates the use of a universal probabilistic programming language for inference on a model without fixed dimension (the number of objects is unknown). Data is simulated from the model and then filtered using a particle filter, within which the delayed sampling heuristic (Murray et al. 2018) automatically yields a Kalman filter for the tracking of each object. It is used as an example in Murray & Schön (2018), in which further details are available.

This package is open source software.

## Getting started

This package requires the Birch.Cairo package.

To build, use:

birch build


To run, use:

./run.sh


## References

1. L.M. Murray and T.B. Schön (2018). Automated learning with a probabilistic programming language: Birch. Annual Reviews in Control 46:29--43. [arxiv]

2. L.M. Murray, D. Lundén, J. Kudlicka, D. Broman and T.B. Schön (2018). Delayed Sampling and Automatic Rao–Blackwellization of Probabilistic Programs. Proceedings of the 21st International Conference on Artificial Intelligence and Statistics (AISTATS).