In this computer vision project we use three different algorithms to track multiple objects in a video simultaneously. CSRT, KCF and MOSSE can all be used in this exaple.
The CSRT model enables channel and spatial reliability. It is very accurate but not that fast. MOSSE on the other hand is a lot faster but lacks accuracy. The MOSSE filter adapts to the changes in the appearance of the object while tracking. Finally, KCF algorithm uses a linear ridge regression model to track an object. It is as accurate as it is fast.
To run the algorithm on sample video, the user must firstly select one or more areas of interest by drawing a bounding box around it with the mouse. Then this area is added to be tracked. Once the selection process is done the algorithm tracks the objects inside that area.
The project was made using
