Skip to content
Snippets Groups Projects
Commit fb959b46 authored by Anjith GEORGE's avatar Anjith GEORGE
Browse files

Merge branch 'ageorge-master-patch-50437' into 'master'

Update README.rst

See merge request !1
parents 94258f2a a9dbbab2
No related branches found
No related tags found
1 merge request!1Update README.rst
Pipeline #65855 failed
...@@ -76,6 +76,43 @@ After, the installation steps to run the experiments are given in the documentat ...@@ -76,6 +76,43 @@ After, the installation steps to run the experiments are given in the documentat
Check the "sphinx/index.html" for details on how to run the experiments. Check the "sphinx/index.html" for details on how to run the experiments.
Running the experiments
-----------------------
The database class provides a standard interface for training, development and evaluation.
Now, using bob package, all the steps can be automated: training, and evaluation.
To run the experiment, use the following command:
.. code-block:: sh
./bin/bob bio pipelines vanilla-biometrics -vvv \
bob.paper.tifs2022_hfr_prepended_domain_transformer.config.mcxfaceprotocols.VIS-THERMAL-split1 \
bob.paper.tifs2022_hfr_prepended_domain_transformer.config.lightning.x_iresnet_PDT -m
.. note::
Note that the first config file specifies the database to be used and the protocol, one can change the config file to
run the experiment for different datasets and protocols. Please inspect the configuration files and pipelines to familiarize the experiments. Please run this job on a GPU machine for faster target.
Running the command should launch the training, and once the training is finished, the evaluation process would
start automatically. At the end of the experiment, the score files will be written to the folder corresponding to the
specific protocol and dataset.
Evaluating results
------------------
To evaluate the score files run the following command.
.. code-block:: python
./bin/python bob/paper/tifs2022_hfr_prepended_domain_transformer/scripts/hface_metrics.py -sf path_to_scores-dev.csv
It should return AUC, EER, Rank-1 accuracy and Verification rates at different FAR points.
Contact Contact
------- -------
......
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment