The participants in the experiments of the mobile scenario have to read
carefully the next submission instructions before they submit their
system:
○ Each participant is required to provide:
- 2 or 3 executables (it depends on the experiment). The inputs
/ outputs of these executables are imposed by the organizer (see the
description below). Furthermore, the form of the executables has to be
either a perl script or a binary obtained by the compilation of C/C++
source code.
- 2 descriptive documents: a theoretical
documentation which contains some general, high-level information about
their algorithms and a technical documentation which describes the form
and the architecture of their executables.
○ The executables will be run by the organizer on a Linux cluster.
○ The participants taking part in one of the four monomodal experiments
(fingerprint, on-line handwritten signature, talking face and face
video sequences) are required to submit three executables:
- Features extraction: This executable aims at pre-processing
the input data file. Then, it extracts the features from the
pre-processed file. It takes as input one data file (from the BioSecure
mobile database) and outputs features files.
- Model creation:
This executable aims at building a client model from extracted
features. It takes as inputs the features files needed to build the
model and outputs the client model file. This executable is required
only for the algorithms using a model.
- Matching: This
executable aims at matching test features and a client model. It takes
as inputs the test features files and the client model file and
provides a score.
It is noticed that the format of the features and model files is not imposed by the organizer.
○ The participants taking part in the multimodal (fusion) evaluation
are required to submit one executable for each scenario (see the page
Mobile Scenario / Protocol):
- Fusion: This executable aims at fusing monomodal scores. It
takes as inputs 3 monomodal scores produced by the
BioSecure monomodal reference systems (signature, face video sequences and fingerprint) and provides a fused score.
○ You can find below a detailed documentation about the submission instructions for the monomodal experiments:
-> submissionInstructionsMono.pdf
○ You can find below a detailed documentation about the submission instructions for the multimodal experiment:
-> submissionInstructionsMulti.pdf
|