If you do not need any 3D integration, you can reference HelloARVideo sample (Android/iOS).
But strictly speaking, that sample still have a little code about 3D (raw OpenGL). The process video cover and moving along with the whole marker is theoretically a 3D process.
In case you need just a trigger (play video but not track the image, full screen play for example), you can remove the 3D part of the sample and it will still work, and use a native video player (instead of draw video frame using OpenGL) would be a better choice.