Simple face recognition in a picture with many persons. The idea of this code is to verify is the subject present in the query image is present in the session image. In this simple example, the query image is Query_6.png and the session image is Session_3.png. The result is shown in detection.png.

• Install Homebrew /usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"
• Re-install boost-python: brew reinstall boost-python --with-python3 --without-python

• pip3 install --upgrade pip
• pip3 install cmake
• cd dlib
• python3 setup.py install
• pip3 install face_recognition
• pip3 install scikit-image
• pip3 install opencv-python
• pip3 install keras
• pip3 install tensorflow

• git clone https://github.com/davisking/dlib.git
• git clone https://github.com/ageitgey/face_recognition
• git clone https://github.com/nyoki-mtl/keras-facenet
• in folder keras-facenet/model create keras
• in folder keras-facenet/model/keras copy this model https://www.dropbox.com/s/o65ow1cl3lcwd1o/facenet_keras.h5?dl=0

In a folder called "facer" copy the following files:

• main.py
• utils.py
• Query_6.png
• Session_3.png

Run main.py (Expected output: query image is detected in session image).

• To understand main.py: edit the file and read the comments, all functions are defined in utils.py
• The demo uses dlib. If you want to run the demo using FaceNet:

1. edit main.py and uncomment line 17 from keras.models import load_model
2. in line 32 put the path of model facenet_keras.h5
3. edit main.py, in line 14 set fr_method = 2
4. run main.py

• Query Image: Query_6.png

• Session Image: Session_3.png

• Is the person in the Query Image present in the Session Image?

[facer] : ----------------------------------------------------------------

[facer] : fd = HOG, fr = Dlib+, sc = CosSim(uninorm=1), th = 0.6

[facer] : loading face recognition model...

[facer] : reading query image Query_6.png...

[facer] : detecting faces in session image Session_3.png...

[facer] : 13 face(s) found in session image Session_3.png

[facer] : finding Query_6.png in session image ...

[facer] : scores

[0.84946395 0.90679934 0.87062143 0.86717102 0.87782153 0.87154528

0.95557544 0.89859898 0.90368649 0.86654181 0.84295712 0.84654535

0.8821119 ]

[facer] : face #6 was detected with score = 0.9555754407241641

References:

• Mery, D.; Mackenney, I.; and Villalobos, E. Student Attendance System in Crowded Classrooms using a Smartphone Camera. In 2019 IEEE Winter Conference on Applications of Computer Vision (WACV2017), 2019.

Note: The students of these pictures gave their permission to be part of this research. The pictures can be used free of charge, for research and educational purposes only. Redistribution and commercial use is prohibited.