I had established stereovision similar to human eyes by connecting 2 webcams together and synchronising them. I moved 2 similar objects in random patterns and traced their path in 3 dimensions, x,y and z axes by stitching the images from both the cameras in each frame together to obtain depth information.
Using the positions obtained, I traced the path of each object in 3 dimensions. Sometimes, one object get occluded by the other. During such cases, I used reverse-jump Markov-chain Monte Carlo algorithm to predict its path.
With the technology available in 2011, I was able to correctly track 2 objects in non real-time successfully for a 2 min video.