Question about disappearing landmark


I’m trying to track a point that (as part of the experiment) disappears halfway through the video. So far I just label this point like I would do other points of interest, and am planning to filter out the spurious points in the csv file. Does this sound ok?

Another question: if I want to track multiple videos (each video has slightly different light conditions), do I need to label frames from different videos? Can you link to the manual where it explains how to do this?

points can absolutely disappear - i.e. only label it when you see it. Then the confidence will be low, and you can threshold the data with pcutoff.

Yes, you should try to label frames from diverse videos. Here are resources:

1 Like