In this homework for CMU's Advanced CV course, I implemented template tracking using the Lucas Kanade tracking method.
A brief description of the method follows
- The algorithm works on two images. One image is the template
$\mathcal{T}$ and one the image$\mathcal{I}$ - Calculate the gradients of either the template or the image based on the variant of the algorithm being run
- Calculate an ideal warp that reduced the error between both images using a gradient descent algorithm