Hi! I have some questions regarding this project: Do you want to stick with OpenCVs alternative (center and tilted features; adaboost of decision trees) or is it ok to explore other alternatives? If you prefer a different (maybe newer) approach, then: How much of the original Viola-Jones object detector you want? The title of this thread is "Face detection" so if you just want face detection there are some papers that extend the original algorithm putting the stress on that topic. If you want to have a Viola-Jones generalistic object detection framework we should stick close to the original formulation (use other features [variations to the integral image computation] and other ensemble algorithms [I've seen SVMs too]). In the case you want a generalistic object detection framework, and don't want to stick with Viola-Jones approach; What do you prefer, efficiency or performace? There are some newer methods like pictorial structures that perform good with articulated objects too like the human body but may not be suited (depending on how many parts and which appearance and deformation parameters you chose) for real time applications. Either way, my current area of research is detection of articulated objects and I've just opened an issue ( *https://github.com/scikit-image/scikit-image/issues/904)* on GitHub to know if you want some work towards that direction. I can implement the generalized distance transform now, and leave for the summer the rest of the code. Or just implement the generalized distance transform now (if you want to implement some recursive, energy minimization algorithm in an efficient way you will use it; plus its a good way to show my skills) and discuss what would be the best approach for this project. Thanks for your time! El jueves, 28 de marzo de 2013 10:35:31 UTC+1, Stefan van der Walt escribió:

Hi everyone

I've been interested in getting face detection into skimage for a while. This morning, Nathan Faggian reminded me that the highly popular Viola-Jones detector is patent encumbered (yes, if you're not careful you can use patented code in packages like OpenCV). However, the following link seems to suggest that we can work around that by training our own classifier with different features:

http://rafaelmizrahi.blogspot.com/2007/02/intel-opencv-face-detection-licens...

If there's any interest in working on this, or if you already have an algorithm available, please get in touch.

Stéfan