Medd

I've made a nifty little patch that works with an Arduino project I'm working on. I click a starting point for cv.jit.track like its help patch (on say the rim of my nose) and it tracks the movement of that point pretty much flawlessly. I've got the XY coordinates controlling a pair of connected servos forming a moving 'head'.

While this functionality is pretty much exactly what I want, the next thing I need is a way of grabbing that point (or any similar point automatically).

I'm not expecting a patch to be made for me, just some advice on how to get a point without clicking on it first. The cv.jit.faces object is sort of performing this action, as is cv.jit.features. Is there a way to easily utilise a combination of these to achieve my goal?

using-cv-jit-features-to-get-cv-jit-track-to-draw-starting-point-or-something

Cracked a dirty little way to this using [cv.jit.faces]:

[toedge] can be used to detect when a face has been picked up, meaning its centre coordinates can be briefly passed onto [cv.jit.track]

This means that a head position can be picked up automatically and continue being tracked. The only downside is that it requires face detection first.

Using cv.jit.features to get cv.jit.track to draw starting point (or something)