OpenCV on OSX using Native APIs
In an attempt to simultaneously learn the OpenCV API as well as those for OSX, I created this small project to recreate the FaceTracker example that accompanies the OpenCV framework. 

This project makes use of the OpenCV framework, as built from CVS (It’s included in the zip below). Moreover, it uses the QTKit to capture video and pipe it into OpenCV for face detection. It seems that doing so gives better performance than using the cv capture methods.

Because of threading issues with the way Quicktime captures video, faces are detected on the main thread, which makes the user interfaces less responsive than it could. I’d appreciate any thoughts, comments and code that might improve UI performance.

Feel free to use this code in your projects. There’s nothing really new here, other than pipe fittings between Apple’s CoreVideo and OpenCV.

Also, if anyone has any insights about using Apple’s Core Image framework as opposed to the vImage methods in the Accelerate framework, please let me know.

Click on the image above to download the XCode project.OpenCV on OSX_files/FaceDetection.zip