DevinBayly
commented
2 years ago three main features of kalman in object tracking
- prediction of object's future position
- reduction of noise due to inaccurate detections
- facilitating the process of associations
** tracks might mean like tracks that a train goes along, like an actual path
Issues arising in object tracking
- Noise is introduced by imperfect background segmentation, this is intended to get corrected by the filter
- Occlusion sometimes occurs
predictions occur followed by corrections which is similar to the update steps mentioned in this resource https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python
prediction alone is used in occlusion instances
configurations options can be tweaked to optimize
Links back to article discussing https://www.mathworks.com/help/vision/ug/motion-based-multiple-object-tracking.html
https://www.mathworks.com/help/vision/ug/using-kalman-filter-for-object-tracking.html
This should help me understand how they are using the values that they get back from the detection to actually make sure that the object's tracked in memory are accurate.