I’d like to use a camera and some sort of tracking technique to determine if a crowd is leaning to the left or to the right. So if 50 people are all facing the camera with arms in the air - I would like to measure which direction they are leaning.
I think it would work if I was able to capture 3 sample images as templates for comparison: sample 1 would be the crowd standing straight, sample 2 would be leaning left and sample 3 would be leaning right.
I could then compare the current frame with the three templates and determine which was the closest match.
Do you think this is doable in VVVV? Any suggestions as to what nodes/techniques to try would be appreciated.
what kind of background? i think i would try to catch the light/dark “ripple” pattern which the arms of not too many people on a light bg create. dunno if it’s possible in your case. the template idea, i’m not sure that will work…
Only if background is bright and illuminated, because clothes in infrared tends to be different then you would expect - white shirt ca nbe dark or black shirt is gonna white. The background can be controlled by you, though.
May be you use Eyesweb for this kind of analysis. There is a node for motion flow. And then just send some data to V4.
Or ive got this idea: if you set your camera quite right over the heads of crowd, then it would only shot arms in air. Then you could catch this (supposed theres a white background) and put it into V4-contour node. there you get the angle of objects. May be this is a way. But could be unstable - if there are more than arms or overlapping arms or or or.
Sounds like I’ll look into Eyesweb - there seems to be very little tutorials and info on how to use it. Anyone have any links or Eyesweb experience/project files that can help me in the right direction?
Or ive got this idea: if you set your camera quite right over the heads of crowd, then it would only shot arms in air. Then you could catch this (supposed theres a white background) and put it into V4-contour node. there you get the angle of objects. May be this is’’
Most likely it would be difficult to have a camera high enough to do this technique. I would like to have the camera in front of the audience.