PAF Plate and Face Recognition

A complete GUI tool to recognize and detect plates and faces in video streaming and files.

Installation - The Lazy Way

1. Install Oracle VirtualBox.
2. Download and import the Virtual Machine I created for you (size 4 GB).
3. Start the system and launch PAF from applications menu.

Installation - The Hard Way

Please run this commands:

sudo apt install cmake
sudo apt install python3-pyqt5    # In case of problems compile from source...
pip3 install numpy
pip3 install dlib                 # In case of problems compile from source...
pip3 install opencv-python

Then you'll need to install OpenALPR, so execute these commands:

sudo apt install libopencv-dev libtesseract-dev git cmake build-essential libleptonica-dev
sudo apt install liblog4cplus-dev libcurl3-dev

Clone the latest code from GitHub

git clone https://github.com/openalpr/openalpr.git
cd openalpr/src
mkdir build
cd build
cmake -DCMAKE_INSTALL_PREFIX:PATH=/usr -DCMAKE_INSTALL_SYSCONFDIR:PATH=/etc ..
make
sudo make install

To test the Alpr library you may try this:

alpr TEST_IMAGE.jpg

Finally, you need to install python bindings:

cd ../bindings/python
sudo python3 setup.py install

Download this repository and execute:

python3 PAF/paf.py

Usage and screenshots

From the home you can configure and control up to 4 cameras.

Each camera can be configured clicking on the corresponding gear symbol. You must enter a URL (rtsp, http or other protocol supported by OpenCV) including credentials e.g. username:myStrongPassword@12.34.56.78/video/live. You may want to detect also unknown faces/plates (i.e. faces and plates that are not a target). You can also select a ROI (region of interest).

From the home, clicking on the rightmost button of each camera you can see all the events. At bottom left there is a button to delete all the events stored with that camera.

You can process video files too. From menu, just select "Process files". The options are similar to the ones above, but you need to set a output destination.

Considerations

1. When using live video the software is not using a buffer. It takes the current frame from the camera. This means that you may lose a face or a plate, because the algorithm usually cannot process 25 frames per seconds on a common machine.

2. Video file processing, instead, processes every frame found in the video file(s). It will use all the CPUs available in parallel to speed up processing.

3. No software is free of bugs. Please report issues!

Enjoy,

Pasquale Lafiosca