face_dlib_descriptors_extractor - Extended-Object-Detection-ROS/wiki_english GitHub Wiki

A utility for creating a file with face vectors for the face identification attribute using the dlib library.

Launch format:

rosrun extended_object_detection face_dlib_descriptors_extractor path_to_face_dir path_to_output_file path_to_shape_predictor path_to_recognition_model

It does not require running ros master.

• path_to_face_dir path to the directory with faces. Image format must be .jpeg, .jpg, .png. Image names will be text labels for faces (no file extension).
• path_to_output_file path where the file with vectors will be saved, which can be fed to the input of the FaceDlib attribute. The file format can be selected .db.
• path_to_shape_predictor path to the shape predictor. You can download from the dlib website http://dlib.net/files/shape_predictor_5_face_landmarks.dat.bz2 and unzip it to the desired folder.
• path_to_recognition_model the path to the trained face recognition model, can be downloaded from the dlib website [http://dlib.net/files/dlib_face_recognition_resnet_model_v1.dat.bz2 ](http://dlib.net/files/dlib_face_recognition_v1.dat.bz2 ](http://dlib.net/files/dlib_face_recognition_v1.dat.bz2) and unziped to the desired folder.

The utility writes information about each file in the specified directory to the console. If the face in the image is correctly recognized, a message will be displayed

[Face_dlib] calculating face descriptors for Moscowsky.jpg...

If no face was recognized in the image, a message will be displayed

[Face_dlib] file Moscowsky.jpg has no faces!

If there are more than one face then

[Face_dlib] file Moscowsky.jpg has more than one faces!

In both cases, this label will not be added to the database.