Skip to content

FaceEngine Structure Overview#

FaceEngine is subdivided into several facilities. Each facility is dedicated to a single function. Below there is a list of all facilities with short descriptions of functionality they provide. Detailed information may be found in corresponding chapters of this handbook.

FaceEngine facility list:

  • Core facility. This facility stores shared low-level FaceEngine types and factories. This facility is responsible for normal functioning of all other facilities by providing settings accessors and common interfaces. The core facility also contains the main FaceEngine root object that is used to create instances of all higher level objects;
  • Face detection facility. This facility is dedicated to object detection. It contains various object detector implementations and factories;
  • Parameter estimation facility. This facility is dedicated to various image parameter estimation, such as blurriness, transformation and so forth. It contains various estimator implementations and factories;
  • Descriptor processing facility. This facility is dedicated to descriptor extraction and matching. The descriptor is a set of features, describing an object, invariant to object transformation, size or other parameters. Descriptor matching allows judging with certain probability whether two objects are the same. This facility contains various descriptor extractors and containers as well as factories, required to produce them.

So, each facility is a set of classes dedicated to some common for them problem domain. Facilities are independent of each other, with several exceptions, like that all higher level facilities depend on the core facility. Interfacility dependencies are thoroughly described in corresponding chapters of this handbook. The actual set of facilities may vary depending on particular FaceEngine distributions as facilities may be licensed and shipped separately.

This handbook describes the very complete FaceEngine distribution, assuming all facilities are available. The facilities are listed in order of increasing complexity. Applying functions from these facilities in this order allows creating a complete face detection, analysis, recognition and matching pipeline with a significant degree of flexibility. The following chapters break down such pipeline in details.