Running baselines in grandtest protocol of WMCA

In the baseline experiments, initially experiments are done separately for all four channels and score fusion is performed. Running experiments for each channel is described below. The feature based baselines implemented are Multi-channel Local Binary Pattern + SVM and Multi-channel Haralick-RDWT-SVM (H-SVM).

For the baselines using MC-Haralick-SVM you need to have mahotas packsge. It is available from conda-forge channel and can be installed as

conda install -c https://conda.anaconda.org/conda-forge mahotas

The steps to reproduce the results for grandtest protocol are listed below.

MC-Haralick-SVM in grandtest protocol

./bin/spoof.py \
protocol-grandtest \
mc-haralick-svm  \
--sub-directory ../grandtest/MC-Haralick-SVM/ -vvv --grid idiap


bob pad metrics -c apcer100 -r attack -e path_to_scores/scores-{dev,eval}

MC-LBP-SVM in grandtest protocol

./bin/spoof.py \
protocol-grandtest \
mclbp-svm  \
--sub-directory ../grandtest/MC-Haralick-SVM/ -vvv --grid idiap


bob pad metrics -c apcer100 -r attack -e path_to_scores/scores-{dev,eval}

BASELINES in Other protocols

The baselines can be repeated in other channels just by changing database configurations and paths as.

protocol-grandtest
protocol-unknown2D
protocol-unknown3D

for grandtest, ‘unknown2d’ and ‘unknown3d’ respectively.

Similarly all the other baselines and protocols can be experimented with, ie by changing the protocols, and the extractor configuration. More details can be found in Bob documentation.

Refer to the accompanying paper for technical details of the methods used.