|
|
OpenCV CamShift Hi everyone, I thought I share some of my work in progress... Green box = Haar detection Blue box = CamShift tracking First: Detect object using haar detection. Second: Pass tracking rectangle over to CamShift tracking patch. Hope you like it.
|
cool. cam shift is interesting. i have thought about trying to implement something similar in qc. that is doing histogram tracking like cam shift.
if you feel up to sharing some source i would be interested in seeing how make your image provider to ipllimage ? or are you going to a mat image.
the reason i am asking is that i have a finger tracker plugin that i have made. I'm getting great results using c++ but when ever i try to implement it in a qc plugin I'm getting really low frame rates.
I'm thinking it has to be a problem with my image provider conversion. basically I'm going rgba down to 8bit 1 channel ipllimage via some filtering. do you mind sharing or explaining a bit about your process ?
this is what I'm trying to do.
the app on the right is sending my finger data and image data to qc via syphon at a much more acceptable frame rate than I'm getting out of a qc plugin. not sure whats it worth but I'm happy to share the finger tracker ;)
Hi Dust,
I would love to share my code with you, I just don't like to share the source to the public just yet as the code is not pretty and work is still in very early stages.
Your project looks really amazing btw, I would love to take a look into your detection algorithm. Maybe you can send me your email address so we can exchange our projects? :)
The image conversation took me a while to make perform good. I tried a few different things but got the best results by simply using the build in color conversation of OpenCV:
First, create a new cvImage image based on the RGBA image buffer and then simply use cvColor to convert it either to BW (e.g. for the hear detection) or RGB.
probably the best email to get me at is....
pix3ldust@gmail.com