From d39402f9c737281cc814f9fa5d4a275a3e82fa38 Mon Sep 17 00:00:00 2001
From: Amir MOHAMMADI <amir.mohammadi@idiap.ch>
Date: Thu, 25 Mar 2021 16:12:26 +0100
Subject: [PATCH] Add SWAN database
---
bob/pad/face/config/swan.py | 32 +++++++++++++
bob/pad/face/database/__init__.py | 2 +
bob/pad/face/database/database.py | 7 ++-
bob/pad/face/database/swan.py | 56 ++++++++++++++++++++++
bob/pad/face/test/test_databases.py | 73 ++++++++++++++++++++++++-----
doc/links.rst | 1 +
setup.py | 4 ++
7 files changed, 161 insertions(+), 14 deletions(-)
create mode 100644 bob/pad/face/config/swan.py
create mode 100644 bob/pad/face/database/swan.py
diff --git a/bob/pad/face/config/swan.py b/bob/pad/face/config/swan.py
new file mode 100644
index 00000000..5f1cad9f
--- /dev/null
+++ b/bob/pad/face/config/swan.py
@@ -0,0 +1,32 @@
+"""The Swan_ Database.
+
+To configure the location of the database on your computer, run::
+
+ bob config set bob.db.swan.directory /path/to/swan/database
+
+
+The Idiap part of the dataset comprises 150 subjects that are captured in six
+different sessions reflecting real-life scenarios of smartphone assisted
+authentication. One of the unique features of this dataset is that it is
+collected in four different geographic locations representing a diverse
+population and ethnicity. Additionally, it also contains a multimodal
+Presentation Attack (PA) or spoofing dataset using low-cost Presentation Attack
+Instruments (PAI) such as print and electronic display attacks. The novel
+acquisition protocols and the diversity of the data subjects collected from
+different geographic locations will allow developing a novel algorithm for
+either unimodal or multimodal biometrics.
+
+PAD protocols are created according to the SWAN-PAD-protocols document.
+Bona-fide session 2 data is split into 3 sets of training, development, and
+evaluation. The bona-fide data from sessions 3,4,5,6 are used for evaluation as
+well. PA samples are randomly split into 3 sets of training, development, and
+evaluation. All the random splits are done 10 times to created 10 different
+protocols. The PAD protocols contain only one type of attacks. For convenience,
+PA_F and PA_V protocols are created for face and voice, respectively which
+contain all the attacks.
+
+.. include:: links.rst
+"""
+from bob.pad.face.database import SwanPadDatabase
+
+database = SwanPadDatabase()
diff --git a/bob/pad/face/database/__init__.py b/bob/pad/face/database/__init__.py
index 1318192d..082597cb 100644
--- a/bob/pad/face/database/__init__.py
+++ b/bob/pad/face/database/__init__.py
@@ -5,6 +5,7 @@ from .casiasurf import CasiaSurfPadDatabase
from .maskattack import MaskAttackPadDatabase
from .replay_attack import ReplayAttackPadDatabase
from .replay_mobile import ReplayMobilePadDatabase
+from .swan import SwanPadDatabase
# gets sphinx autodoc done right - don't remove it
@@ -30,6 +31,7 @@ __appropriate__(
MaskAttackPadDatabase,
CasiaSurfPadDatabase,
CasiaFasdPadDatabase,
+ SwanPadDatabase,
)
__all__ = [_ for _ in dir() if not _.startswith('_')]
diff --git a/bob/pad/face/database/database.py b/bob/pad/face/database/database.py
index b6450063..d4323f00 100644
--- a/bob/pad/face/database/database.py
+++ b/bob/pad/face/database/database.py
@@ -17,6 +17,7 @@ def delayed_video_load(
max_number_of_frames=None,
step_size=None,
get_transform=None,
+ keep_extension_for_annotation=False,
):
if get_transform is None:
def get_transform(x):
@@ -37,7 +38,9 @@ def delayed_video_load(
)
annotations, delayed_attributes = None, None
if annotation_directory:
- path = os.path.splitext(sample.filename)[0]
+ path = sample.filename
+ if not keep_extension_for_annotation:
+ path = os.path.splitext(sample.filename)[0]
delayed_annotations = partial(
read_annotation_file,
file_name=f"{annotation_directory}:{path}.json",
@@ -63,6 +66,7 @@ def VideoPadSample(
max_number_of_frames=None,
step_size=None,
get_transform=None,
+ keep_extension_for_annotation=False,
):
return FunctionTransformer(
delayed_video_load,
@@ -74,6 +78,7 @@ def VideoPadSample(
max_number_of_frames=max_number_of_frames,
step_size=step_size,
get_transform=get_transform,
+ keep_extension_for_annotation=keep_extension_for_annotation,
),
)
diff --git a/bob/pad/face/database/swan.py b/bob/pad/face/database/swan.py
new file mode 100644
index 00000000..5d0bb75e
--- /dev/null
+++ b/bob/pad/face/database/swan.py
@@ -0,0 +1,56 @@
+import logging
+
+from bob.extension import rc
+from bob.extension.download import get_file
+from bob.pad.base.database import FileListPadDatabase
+from bob.pad.face.database import VideoPadSample
+
+logger = logging.getLogger(__name__)
+
+
+def SwanPadDatabase(
+ protocol="pad_p2_face_f1",
+ selection_style=None,
+ max_number_of_frames=None,
+ step_size=None,
+ annotation_directory=None,
+ annotation_type=None,
+ fixed_positions=None,
+ **kwargs,
+):
+ name = "pad-face-swan.tar.gz"
+ dataset_protocols_path = get_file(
+ name,
+ [f"http://www.idiap.ch/software/bob/data/bob/bob.pad.face/{name}"],
+ cache_subdir="datasets",
+ # file_hash="a8e31cc3",protocols
+ )
+
+ if annotation_directory is None:
+ name = "annotations-swan-mtcnn.tar.xz"
+ annotation_directory = get_file(
+ name,
+ [f"http://www.idiap.ch/software/bob/data/bob/bob.pad.face/{name}"],
+ cache_subdir="annotations",
+ # file_hash="3ecbfa3c",
+ )
+ annotation_type = "eyes-center"
+
+ transformer = VideoPadSample(
+ original_directory=rc.get("bob.db.swan.directory"),
+ annotation_directory=annotation_directory,
+ selection_style=selection_style,
+ max_number_of_frames=max_number_of_frames,
+ step_size=step_size,
+ keep_extension_for_annotation=True,
+ )
+
+ database = FileListPadDatabase(
+ dataset_protocols_path,
+ protocol,
+ transformer=transformer,
+ **kwargs,
+ )
+ database.annotation_type = annotation_type
+ database.fixed_positions = fixed_positions
+ return database
diff --git a/bob/pad/face/test/test_databases.py b/bob/pad/face/test/test_databases.py
index bb4d5129..c1162b24 100644
--- a/bob/pad/face/test/test_databases.py
+++ b/bob/pad/face/test/test_databases.py
@@ -15,17 +15,26 @@ def test_replayattack():
package_prefix="bob.pad.",
)
- assert database.protocols() == ['digitalphoto', 'grandtest', 'highdef', 'mobile', 'photo', 'print', 'smalltest', 'video']
+ assert database.protocols() == [
+ "digitalphoto",
+ "grandtest",
+ "highdef",
+ "mobile",
+ "photo",
+ "print",
+ "smalltest",
+ "video",
+ ]
assert database.groups() == ["dev", "eval", "train"]
assert len(database.samples(groups=["train", "dev", "eval"])) == 1200
assert len(database.samples(groups=["train", "dev"])) == 720
assert len(database.samples(groups=["train"])) == 360
- assert len(database.samples(groups=["train", "dev", "eval"])) == 1200
assert (
len(database.samples(groups=["train", "dev", "eval"], purposes="real")) == 200
)
assert (
- len(database.samples(groups=["train", "dev", "eval"], purposes="attack")) == 1000
+ len(database.samples(groups=["train", "dev", "eval"], purposes="attack"))
+ == 1000
)
sample = database.sort(database.samples())[0]
@@ -45,7 +54,6 @@ def test_replayattack():
raise SkipTest(e)
-
def test_replaymobile():
database = bob.bio.base.load_resource(
"replay-mobile",
@@ -59,7 +67,6 @@ def test_replaymobile():
assert len(database.samples(groups=["train", "dev", "eval"])) == 1030
assert len(database.samples(groups=["train", "dev"])) == 728
assert len(database.samples(groups=["train"])) == 312
- assert len(database.samples(groups=["train", "dev", "eval"])) == 1030
assert (
len(database.samples(groups=["train", "dev", "eval"], purposes="real")) == 390
)
@@ -94,8 +101,7 @@ def test_maskattack():
)
# all real sequences: 2 sessions, 5 recordings for 17 individuals
assert (
- len(maskattack.samples(groups=["train", "dev", "eval"], purposes="real"))
- == 170
+ len(maskattack.samples(groups=["train", "dev", "eval"], purposes="real")) == 170
)
# all attacks: 1 session, 5 recordings for 17 individuals
assert (
@@ -139,6 +145,7 @@ def test_maskattack():
# == 57710
# )
+
def test_casiasurf_color_protocol():
casiasurf = bob.bio.base.load_resource(
"casiasurf-color",
@@ -150,18 +157,14 @@ def test_casiasurf_color_protocol():
assert len(casiasurf.samples(groups=["train"], purposes="attack")) == 20324
assert len(casiasurf.samples(groups=("dev",), purposes=("real",))) == 2994
assert len(casiasurf.samples(groups=("dev",), purposes=("attack",))) == 6614
- assert (
- len(casiasurf.samples(groups=("dev",), purposes=("real", "attack"))) == 9608
- )
+ assert len(casiasurf.samples(groups=("dev",), purposes=("real", "attack"))) == 9608
assert len(casiasurf.samples(groups=("eval",), purposes=("real",))) == 17458
assert len(casiasurf.samples(groups=("eval",), purposes=("attack",))) == 40252
assert (
- len(casiasurf.samples(groups=("eval",), purposes=("real", "attack")))
- == 57710
+ len(casiasurf.samples(groups=("eval",), purposes=("real", "attack"))) == 57710
)
-
def test_casia_fasd():
casia_fasd = bob.bio.base.load_resource(
"casiafasd",
@@ -177,3 +180,47 @@ def test_casia_fasd():
assert len(casia_fasd.samples(groups="train")) == 180
assert len(casia_fasd.samples(groups="dev")) == 60
assert len(casia_fasd.samples(groups="eval")) == 360
+
+
+def test_swan():
+ database = bob.bio.base.load_resource(
+ "swan",
+ "database",
+ preferred_package="bob.pad.face",
+ package_prefix="bob.pad.",
+ )
+
+ assert database.protocols() == [
+ "pad_p2_face_f1",
+ "pad_p2_face_f2",
+ "pad_p2_face_f3",
+ "pad_p2_face_f4",
+ "pad_p2_face_f5",
+ ]
+ assert database.groups() == ["dev", "eval", "train"]
+ assert len(database.samples(groups=["train", "dev", "eval"])) == 5802
+ assert len(database.samples(groups=["train", "dev"])) == 2803
+ assert len(database.samples(groups=["train"])) == 2001
+ assert (
+ len(database.samples(groups=["train", "dev", "eval"], purposes="real")) == 3300
+ )
+ assert (
+ len(database.samples(groups=["train", "dev", "eval"], purposes="attack"))
+ == 2502
+ )
+
+ sample = database.sort(database.samples())[0]
+ try:
+ assert dict(sample.annotations["0"]) == {
+ "bottomright": [849, 564],
+ "leye": [511, 453],
+ "mouthleft": [709, 271],
+ "mouthright": [711, 445],
+ "nose": [590, 357],
+ "reye": [510, 265],
+ "topleft": [301, 169],
+ }
+ assert sample.data.shape == (20, 3, 720, 1280)
+ assert sample.data[0][0, 0, 0] == 87
+ except RuntimeError as e:
+ raise SkipTest(e)
diff --git a/doc/links.rst b/doc/links.rst
index cd652a07..09f5c2d7 100644
--- a/doc/links.rst
+++ b/doc/links.rst
@@ -14,3 +14,4 @@
.. _dependencies: https://gitlab.idiap.ch/bob/bob/wikis/Dependencies
.. _MIFS: http://www.antitza.com/makeup-datasets.html
.. _CELEBA: http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html
+.. _Swan: https://www.idiap.ch/dataset/swan
diff --git a/setup.py b/setup.py
index 088ed775..10450c19 100644
--- a/setup.py
+++ b/setup.py
@@ -62,6 +62,7 @@ setup(
"maskattack = bob.pad.face.config.maskattack:database",
"casiasurf-color = bob.pad.face.config.casiasurf_color:database",
"casiasurf = bob.pad.face.config.casiasurf:database",
+ "swan = bob.pad.face.config.swan:database",
],
# registered configurations:
"bob.pad.config": [
@@ -70,6 +71,9 @@ setup(
"replay-mobile = bob.pad.face.config.replay_mobile",
"casiafasd = bob.pad.face.config.casiafasd",
"maskattack = bob.pad.face.config.maskattack",
+ "casiasurf-color = bob.pad.face.config.casiasurf_color",
+ "casiasurf = bob.pad.face.config.casiasurf",
+ "swan = bob.pad.face.config.swan",
# LBPs
"lbp = bob.pad.face.config.lbp_64",
# quality measure
--
GitLab