IJB-C Database Interface for Bob¶
This package is part of the signal-processing and machine learning toolbox Bob. This package contains an interface for the evaluation protocols of the IARPA Janus Benchmark C (IJB-C) database and does not contain the original image data for the database. The original data should be obtained using the link above.
The IJB-C database is a mixture of frontal and non-frontal images and videos (provided as single frames) from 1845 different identities. In many of the images and video frames, there are several people visible, but only the ones that are annotated with a bounding box should be taken into consideration. For both model enrollment as well as for probing, images and video frames of one person are combined into so-called Templates. For some of the protocols, probe templates are also generated from raw video data.
This package implements the database interface including all its particularities:
- First, it implements the FileSet protocol, since for some probes, several files (a mixture of images and/or video frames) are defined. In the Database.object_sets() function, FileSet objects are only returned for probe purposes.
- Second, some images contain several identities. Therefore, every physical image file can be stored in several File objects. Also, the File.make_path() function can create two different styles of file names: the original file name (to read original images), or a unique filename (to define a unique name for each extracted face).
Follow our installation instructions. Then, using the Python interpreter provided by the distribution and buildout this package: