search

OpenSourcePhysics / tracker

Video analysis and modeling tool built on the Open Source Physics framework
GNU General Public License v3.0
206 stars 52 forks source link

[feature request] Autostop of autotracker #16

Open nickkolok opened 5 years ago

nickkolok commented 5 years ago

It would be very good if the autotracker had an option "Autostop" with two parameters: frame count n and precisions epsilon_x and epsilon_y. Let's say that a point is not moving during n frames iff there exist intervals of length epsilon_x and epsilon_y such that the point x and y coords are in this intervals for all the n frames.

Autostop happens when firstly the point is moving and then stops (if the point isn'tmoving for first n frames, then tthe autosearch process continues.

This feture could be useful in certain cases, i.e. when aone video consists of three parts: a) The object of interest isnot moving, preparation happens b) The object is moving, the desired phenomenon happens c)The object isn't moving, but the camera is still recording

Such cases sometimes really take place in real activirties, but in general I would say that the priority of this request is low.

dobrown commented 5 years ago

I think this would be very rarely used by students or even most researchers since the autotracker is not normally left to run unattended. Better to just set reasonable start and end frames IMHO.

nickkolok commented 5 years ago

autotracker is not normally left to run unattended

Oh, it is. Our team, which participates in IPT (International Physicists' Tournament), often needs to autotrack tens (e.g. 60 - on last Monday) videos per evening as fast as possible. So, we run some Tracker windows with autotracker in parallel and we want to automatize the process as much as possible.

dobrown commented 5 years ago

OK, but setting the in and out frames is very fast, especially if the out frame has only to be any time after the object has come to rest... Bottom line: I'm not planning to implement autostop, sorry.