2D trackers come in two flavors, automatic and manual. This distinction has nothing to do with how they work, as all trackers do exactly the same thing, record the pattern in one frame, search for it on the next frame and therefore record a 2D trail of the specific image feature. These differences are the following: How were they created? Who chooses the original placement? And what happens when they lose the pattern? As far as the creation and original placement are concerned, automatic trackers are randomly generated on a frame of our shot, and any of them of them that happen to be on a high contrast are kept, while the others are deleted. With that original position as a start, each tracker tries to follow the feature underneath it for as long as it can. On the other hand, manual trackers are placed by the user on parts of the image we think are good candidates for a tracker to follow. As far as what happens after the pattern is lost, the automatic trackers just stop recording 2D positions, whereas the manual trackers expect the user to realign them, if the feature is still available, or tell them to stop recording. Automatic tracking is a very straightforward process and comes with some interesting attributes for creating and filtering. On the other hand, manual tracking is much more labor intensive. Since the manual tracking process has a semi-automatic component, it's referred to as supervised tracking as well. During this series, I may refer to it as either manual or supervised in various occasions, but I essentially mean exactly the same thing