Based on unmarked AR: The representative works are PTAM/M, Mixare, which will be the future development direction of AR
Tracking technology can be roughly divided into two categories, one is Feature Based Tracking, for example, by tracking the feature points and feature edges extracted from the input image to complete the 3D motion estimation and tracking of the target.
The other type is Template Based Tracking, that is, tracking is achieved by matching the grayscale image of part of the input image.
Feature Based Tracking The representative work of Feature Based Tracking is PTAM of the Robotics Institute of the Department of Engineering, Oxford University, see Georg Klein's homepage at the university http://www.robots.ox.ac.uk/~gk/PTAM/ You can download the learning source code.
Template Based Tracking The representative work of Template Based Tracking is the work of the Computer Vision Laboratory of EPFL, see the series of real-time tracking articles on Vincent Lepetit's homepage http://cvlab.epfl.ch/~lepetit/
It is also worth downloading the source code on the homepage of Vincent's collaborator Stefan Hintersser http://campar.in.tum.de/Main/StefanHinterstoisser , most of which are based on OpenCV.