How do I integrate SURF algorithm with particle filter tracking?

See more:

OpenCV

tracking

I'm new to the image processing field. I'm currently trying to detect multiple persons in a video using the SURF algorithm with SVM training for detection, and the particle filter for tracking. So far, I used the bag-of-features to build a vocabulary and cluster keypoints from a set of images (for both persons and nonpersons) and used two SVMs to classify if the test data are persons or not. Now that the classification is done, I'm planning to use the SURF keypoints extracted from the test data for the particle filter.

Basically, I want to justify the SURF keypoints to suit the particle filter input, feed these points into the particle filter algorithm, and then start tracking the targets and drawing circles representing the particles at the same time.

How do I do this? :( I'm trying to study the code provided here https://github.com/NewProggie/Particle-Filter[^] but I kinda got lost because it AdaBoost, and of course, the particles are generated randomly. Thank you very much!