This the open source code used for the paper Lalauze-Pol R, Jouen F. Facial Growth in Children from 1 Month to 7 Years: A Biometric Approach by Image Processing. GSL J Pediatr. 2020; 1:105.
#September 2026 New versions for both x86 (macOS and Winows) and ARM-64 (macOS Tahoe 26.7) processors!
First of all you need to install Red Language. Download a recent or the latest version (Automated builds, master branch). See Red Programming Language.
Face also uses redCV library which is included in redCV/libs directory. New redCV libs (2026) are included in redCV/libs/ directory
Face software also requires python 3: python >= 3.8 .
In these vesrsions we use Python 3.10.20.
Face software requires DLib c++ library for facial landmarks recognition.
Dlib shape_predictor_68_face_landmarks.dat is in the Face directory.
In most cases, required python libraries are still installed.
For installing dlib for Python just do pip install dlib or pip3 install dlib in your terminal.
Installation can be a little bit complicated with Windows.
You need to install cMake. Have a look here cmake install .
You can also need to install Visual Studio build tools from here.
In Visual Studio go to the Individual Components tab, Visual C++ Tools for Cmake, and check the checkbox under the "Compilers, build tools and runtimes" section. Then save the modifications.
If all is correct, pip install dlib should be sufficient for installing python dlib version.
Whatever the version modify :
appDir: to-red-file rejoin [home "/Programmation/Red/Face/"] to adapt to your path
Compilation is very easy: just cd to the face-main directory and type red -c face.red. For Windows users, use WindowsPowerShell as terminal.
In case of problems, type red -u -c face.red to upgrade runtime library.
Both versions can be compilated with Red-GUI or Red-CLI. Red-CLI is useful for debugging code if necessary.
Face processes 2-D images of faces for measuring facial heights, ratios and surfaces. You'll find in /publication directory, our paper which gives all details about the used method.
Load a source image e.g. /images/face1.png, and modify angle if necessary. Then save the modified image. This creates a copy of the source image, e.g. /images/face1C.png. Now you can detect the landmarks with DLib network and compute facial heights and surfaces.
You can also load an image as a processed image. Do not use the copy image file name, but the source image file name. The mapping betwen both images is automatic. Now you can find landmarks and compute values.
Computed values can be copied to an excel sheet: Classeur1.xlsx.