Face Detector

Input image

voyager crew

Detector output

format:
/path/input_image.pgm
nb_of_faces
x₁y₁size₁part₁ side₁
      :
      :

output example:
voyager2.pgm.fmax

Image with displayed regions

crew's faces detected

Parameters defining a face

x, y - coordinates of the face center with respect to upper left corner of the image.
size - size in pixels of a box bounding eyes and mouth
confidence - detection confidence for thresholding
part - body part - 0 for face
side - side of the face - 1 for frontal face

Linux binaries	Example of use
face_det.tar.gz	To run the detector on example images. The input images are images/.pgm and the output files are .pgm.fmax and face_.pgm: `prompt>./test.sh`. to detect faces in an image:* `prompt>./det_face.ln -data ./BOOST_FFACE/ -D1 -i ./images/voyager2.pgm -suff .pgm -s1 20 -o face_out.fmax` to draw squares on detected faces with confidence threshold set to -50: `prompt>./det_face.ln -DR -i ./face_out.fmax -t2 -50`
frontal.dat.gz	New face detector. To run the detector on example images download `compute_descriptors` from Feature Detection Code. The input images can be in png, JPG etc formats and the output files are defined as in Feature Detection Code: `prompt>./test.sh`. to detect faces in an image:* `prompt>./compute_descriptors.ln -fface frontal.dat -i voyager2.jpg -o2 voyager.fmax` to detect and draw circles on detected faces: `prompt>./compute_descriptors.ln -fface frontal.dat -i voyager2.jpg -o2 voyager.fmax -DC`
	Related papers
	Mikolajczyk,K. , Schmid, C. and Zisserman, A., Human detection based on a probabilistic assembly of robust part detectors. ECCV 2004. PDF Kalal, Z. , Matas, J. and Mikolajczyk, K., Weighted sampling for large scale boosting. BMVC 2008.PDF