The purpose of this project is to allow a connected mirror to serve as a personalized information point for trade fair/congress visitors.
To make this point of information personalized, we have chosen to deal with facial recognition (open source OpenCV project with facial_recognition lib) which allows us to automatically connect a visitor to our CMS solution. An acquisition module is also available on another repo, it allows to take the reference image/photo for facial recognition.
For a question of optimization, 2 ubuntu services were created on the raspberry of our project.
The flask server (identification/server.py) runs on the systemd/user because it must not be initiated by root & only by the main user (here it's grav), it is this service that will load most of the libraries.
The detection program (presence_detector.py) runs in systemd/system since it needs to be initiated by root to have access to the GPIO.
All the links are absolute because of this services & because i don't use python environment for libs install
sudo apt install libopencv-dev python3-opencv
python3 pip install
or
wget https://bootstrap.pypa.io/get-pip.py
sudo python3 get-pip.py && rm get-pip.py
pip install face-recognition
pip install python-dotenv
pip install paramiko
pip install scp
pip install numpy
sudo apt install -y cmake
pip install dlib // Not the fastest way to install dlib
Autopy need to read documentation for install // Autopy lib need small correction on MAKEFILE
Mediapipe compil for ARM64 - Instructions
... & Maybe more
L42Project (Facial Recognition [FR]):
Mohammadst99 repo:
SR04 Instructions:
OpenCV function imshow segfault. GITHUB ISSUE
RESOLUTION: remove opencv installation with apt & install the github version.
FOR OPENCV 4.5.5 version from github official repo you need to use cmake with a flag:
cmake ../ -D PYTHON3_PACKAGES_PATH=/usr/lib/python3/dist-packages