The intended purpose of this unofficial fork/branch is to make OFIQ more easily/efficiently usable from Python, and also to expose additional OFIQ_LIB::Session data.
Note that this is currently mainly meant for internal use, so the documentation is rather minimal and there are some caveats regarding the dependencies (see blow).
Quick start: Build OFIQ as per usual, then see the documentation at the top of python/ofiq_zmq.py.
Python dependencies: The Python side (python/ofiq_zmq.py) currently requires an installation of the FIQA Toolkit "fiqat".
This requirement could be removed if desired, as it is mainly used to handle image loading.
Besides that it requires the packages numpy (tested version 1.23.5) and pyzmq (tested version 25.0.2).
The tested Python version is 3.9.16.
Added C++ dependencies: The OFIQ build now requires ZeroMQ (specifically the C version, i.e. "libzmq"), which should however be handled automatically by the Conan package manager (i.e. zeromq/4.3.5 has been added to the conan/conanfile.txt).
But note that the build now also requires C++20 (CMAKE_CXX_STANDARD 20), albeit only for std::endian in the new OFIQlib/src/OFIQ_zmq_app.cpp file.
Details & extensibility to other languages: To facilitate the Python usage, this adds a C++ executable target with the file OFIQlib/src/OFIQ_zmq_app.cpp, which acts as a ZeroMQ server (ZMQ_REP) through which OFIQ can be used.
The single added client (ZMQ_REQ) is the Python module python/ofiq_zmq.py, which controls the start/stop of the OFIQ_zmq_app executable.
Technically other language adapters could use the OFIQ_zmq_app executable as well, but there currently are no plans to implement any.
C++ file changes: Some OFIQ C++ files have been extended to expose OFIQ_LIB::Session data.
As a result, OpenCV is now required as a dependency for applications that use this modified OFIQ version (e.g. OFIQSampleApp.cpp which otherwise doesn't need OpenCV). This could be fixed if required, but there currently is no plan to do so, as this is meant as a modified standalone OFIQ fork/branch that allows Python usage.
License: Like the majority of the OFIQ code, the new Python file (python/ofiq_zmq.py) and C++ file (OFIQlib/src/OFIQ_zmq_app.cpp) use the MIT License (see the file headers).
Miscellaneous: One scripts/build.sh variant has been added (build__only_build.sh), but that's an unimportant helper for C++ development purposes. A .gitignore file has also been added.
The remainder of this README is the original content (except with nested headers).
The OFIQ (Open Source Face Image Quality) is a software library for computing quality aspects of a facial image. OFIQ is written in the C/C++ programming language. OFIQ is the reference implementation for the ISO/IEC 29794-5 international standard; see https://bsi.bund.de/dok/OFIQ-e.
Before using OFIQ or distributing parts of OFIQ one should have a look on OFIQ's license and the license of its dependencies: LICENSE.md
For a tutorial on how to compile and operate OFIQ, see here.
A full documentation of OFIQ including compilation, configuration and a comprehensive doxygen documentation of the C/C++ API is contained in the reference manual: see doc/refman.pdf.