Classic algorithms for tracking uncooperative moving objects are often computationally expensive, even with the most powerful machines in the consumer market. The challenge comes from implementing the tracker’s calculation inside a database management system.

In this paper, the authors propose optimizations that rely on incremental updates. Their approach reduces the runtime of probabilistic multiple hypothesis tracking. The computational complexity of this approach is linear in terms of the amount of data processed. Since the implementation is SQL-based, the solution is also easily modifiable and extensible. The theory is well explained in the paper, and supported by concrete data. Overall, I thoroughly enjoyed reading the paper.