oneapi-antispoofing

Certiface AntiSpoofing use oneAPI for fast decode video for perform liveness detection with inference. The system is capable of spotting fake faces and performing anti-face spoofing in face recognition systems. The user could try to hold up a your photo. Maybe they even have a photo or video on their smartphone (obtained in facebook, for example)

This project/solution use harness heterogeneous computing architecture including CPUs and GPUs from servers to notebooks. The software tools such as oneVPL, computer vision techniques with openCV and Deep Learning technologies based on Intel features.​ ​

Methodology / Approach

Using deep learning for detect photo with fast decoder h264/h265 format video. Technologies Used:

• oneAPI
• openCV
• openVINO

Technologies Used: oneAPI and openCV

Installation instructions:

Under construction

Below, a quick description of how to manually install the oneapi-antispoofing project.

Prepare ambient

Install the openCV library and set up the oneapi development environment as in the example below.

$ source /opt/intel/oneapi/setvars.sh 
:: initializing oneAPI environment ...
:: dnnl -- latest
:: advisor -- latest
:: dpcpp-ct -- latest
:: dev-utilities -- latest
:: ipp -- latest
:: ccl -- latest
:: compiler -- latest
:: ippcp -- latest
:: daal -- latest
:: debugger -- latest
:: mpi -- latest
:: intelpython -- latest
:: tbb -- latest
:: vpl -- latest
:: vtune -- latest
:: mkl -- latest
:: oneAPI environment initialized ::

Clone and build the project

Clone the repository at desired location:

$ git clone https://github.com/cabelo/oneapi-antispoofing
$ cd oneapi-antispoofing
$ mkdir build
$ cd build
$ cmake ..
$ make

Run the example

Below, how to run the project. Enter in build folder and execute the commands:

$ wget https://service.assuntonerd.com.br/downloads/antispoofing.weights
$ wget https://service.assuntonerd.com.br/downloads/antispoofing.cfg
$ make run

Or to test fraud, use the command below:

$ ./antispoofing ../content/fraud-video.h265 y

To test real case, use the command below:

$ ./antispoofing ../content/real-video.h265 y

To Do

• First version
• Publish in github
• Teste and port (if necessary) to oneAPI Beta10
• Teste and port (if necessary) to oneAPI 2023
• Teste and port to oneVPL 2023
• Create webserver REST

The final result

Bellow an example running in my machine with Linux OpenSUSE Leap 15.2, 15.3 and 15.4.

contact : Alessandro de Oliveira Faria (A.K.A.CABELO) [email protected]

Thank you Sujata Tibewala and Emma Mai.