A pure Python implementation of Random Forests specifically developed for face detection purposes.
Random forests are an ensemble learning method for classification (and regression) that operate by constructing a multitude of decision trees at training time and outputting the class that is the mode of the classes output by individual trees.
We use the CBCL dataset to train a random forest for face detection.
python trainforest.py datasets/CBCLfaces.pkl python detectfaces.py forests/rf_t20_d-1_r200_s5.pkl images/astro.png