How to hold an object so that the neuron does not lose it?

Question

OpenCV

Finds an object, but loses it, it is necessary that the coordinates of the object be continuously read into the array, that is, it is necessary that the found object be held

Answer 1

If the neuron lost the object, it means that it could not recognize it, which means that it will not be able to determine the coordinates either.
Undertrained network?
The option is to try to use the knowledge from the previous frame, on which the neuron was able to select an object, with some more oaky algorithm, google the opencv motion detector, taking as a basis a piece of the image or rather its parameters, i.e. filtering the frame, select the filters for the task by itself, for example, a simple image differential from the frame where the object was last seen and all subsequent ones, where the object moved will be the maximum, you take its center (you need to determine in advance the center for the desired object using the same algorithm to subtract it, since the coordinates from the neuron you can have the upper left corner)
worst of all, if the camera can also move / turn, or an unnecessary background can move, then you need an algorithm not for the frame difference, but the one that is used to determine the camera shake (for points on the screen, a motion vector will be given, those that the object had from past frames will give a correction for the old coordinates).