Merge pull request #4 from lyt910522/master

add frost metric

examples/torch_frost_example.py

@@ -0,0 +1,75 @@
+""" Test case for Torch """
+
+from __future__ import absolute_import
+
+import torch
+import torchvision.models as models
+import numpy as np
+from perceptron.models.classification.pytorch import PyTorchModel
+from perceptron.utils.image import imagenet_example
+from perceptron.benchmarks.frost import FrostMetric
+from perceptron.utils.criteria.classification import Misclassification
+from perceptron.utils.tools import plot_image
+from perceptron.utils.tools import bcolors
+
+# instantiate the model
+resnet18 = models.resnet18(pretrained=True).eval()
+if torch.cuda.is_available():
+    resnet18 = resnet18.cuda()
+
+# initialize the PyTorchModel
+mean = np.array([0.485, 0.456, 0.406]).reshape((3, 1, 1))
+std = np.array([0.229, 0.224, 0.225]).reshape((3, 1, 1))
+fmodel = PyTorchModel(
+    resnet18, bounds=(0, 1), num_classes=1000, preprocessing=(mean, std))
+
+# get source image and print the predicted label
+image, _ = imagenet_example(data_format='channels_first')
+image = image / 255.  # because our model expects values in [0, 1]
+
+# set the type of noise which will used to generate the adversarial examples
+metric = FrostMetric(fmodel, criterion=Misclassification())
+
+# set the label as the predicted one
+label = np.argmax(fmodel.predictions(image))
+
+print(bcolors.BOLD + 'Process start' + bcolors.ENDC)
+# set 'unpack' as false so we can access the detailed info of adversary
+adversary = metric(image, label, scenario=5, verify=True, unpack=False)
+print(bcolors.BOLD + 'Process finished' + bcolors.ENDC)
+
+if adversary.image is None:
+    print(
+        bcolors.WARNING +
+        'Warning: Cannot find an adversary!' +
+        bcolors.ENDC)
+    exit(-1)
+
+###################  print summary info  #####################################
+
+keywords = ['PyTorch', 'ResNet18', 'Misclassification', 'Frost']
+
+true_label = np.argmax(fmodel.predictions(image))
+fake_label = np.argmax(fmodel.predictions(adversary.image))
+
+# interpret the label as human language
+with open('perceptron/utils/labels.txt') as info:
+    imagenet_dict = eval(info.read())
+
+print(bcolors.HEADER + bcolors.UNDERLINE + 'Summary:' + bcolors.ENDC)
+print('Configuration:' + bcolors.CYAN + ' --framework %s '
+      '--model %s --criterion %s '
+      '--metric %s' % tuple(keywords) + bcolors.ENDC)
+print('The predicted label of original image is '
+      + bcolors.GREEN + imagenet_dict[true_label] + bcolors.ENDC)
+print('The predicted label of adversary image is '
+      + bcolors.RED + imagenet_dict[fake_label] + bcolors.ENDC)
+print('Minimum perturbation required: %s' % bcolors.BLUE
+      + str(adversary.distance) + bcolors.ENDC)
+print('Verifiable bound: %s' % bcolors.BLUE
+      + str(adversary.verifiable_bounds) + bcolors.ENDC)
+print('\n')
+
+plot_image(adversary,
+           title=', '.join(keywords),
+           figname='examples/images/%s.png' % '_'.join(keywords))

perceptron/benchmarks/frost.py

@@ -0,0 +1,112 @@
+# Copyright 2019 Baidu Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Metric that tests models against frost variations."""
+
+import numpy as np
+from tqdm import tqdm
+from collections import Iterable
+from .base import Metric
+from .base import call_decorator
+from PIL import Image
+import warnings
+
+
+class FrostMetric(Metric):
+    """Metric that tests models against frost variations."""
+
+    @call_decorator
+    def __call__(self, adv, scenario=5, annotation=None, unpack=True,
+                 abort_early=True, verify=False, epsilons=1000):
+        """Change the frost of the image until it is misclassified.
+
+        Parameters
+        ----------
+        adv : `numpy.ndarray`
+            The original, unperturbed input as a `numpy.ndarray`.
+        scenario : int or PIL.Image
+            Choice of frost backgrounds.
+        annotation : int
+            The reference label of the original input. Must be passed
+            if `a` is a `numpy.ndarray`.
+        unpack : bool
+            If true, returns the adversarial input, otherwise returns
+            the Adversarial object.
+        abort_early : bool
+            If true, returns when got first adversarial, otherwise
+            returns when all the iterations are finished.
+        verify : bool
+            If True, return verifiable bound.
+        epsilons : int or Iterable[float]
+            Either Iterable of contrast levels or number of brightness
+            factors between 1 and 0 that should be tried. Epsilons are
+            one minus the brightness factor. Epsilons are not used if
+            verify = True.
+
+        """
+
+        if verify is True:
+            warnings.warn('epsilon is not used in verification mode '
+                          'and abort_early is set to True.')
+
+        a = adv
+        del adv
+        del annotation
+        del unpack
+
+        image = a.original_image
+        min_, max_ = a.bounds()
+        axis = a.channel_axis(batch=False)
+        hw = [image.shape[i] for i in range(image.ndim) if i != axis]
+        img_height, img_width = hw
+
+        if not isinstance(epsilons, Iterable):
+            epsilons = np.linspace(0, 1, num=epsilons)[1:]
+        else:
+            epsilons = epsilons
+
+        if isinstance(scenario, Image.Image):
+            frost_img_pil = scenario
+        elif isinstance(scenario, int):
+            frost_img_pil = Image.open(
+                'perceptron/utils/images/frost{0}.png'.format(scenario))
+        else:
+            raise ValueError(
+                'scenatiro has to be eigher int or PIL.Image.Image')
+
+        frost_img = np.array(
+            frost_img_pil.convert('RGB').resize(
+                (img_width, img_height))).astype(
+            np.float32) / 255.
+        frost_img = frost_img * max_
+        if(axis == 0):
+            frost_img = np.transpose(frost_img, (2, 0, 1))
+
+        cc0 = [1.0, 0.5]
+        cc1 = [0.3, 0.8]
+        for _, epsilon in enumerate(tqdm(epsilons)):
+            p0 = cc0[0] + epsilon * (cc0[1] - cc0[0])
+            p1 = cc1[0] + epsilon * (cc1[1] - cc1[0])
+            perturbed = image * p0 + frost_img * p1
+            perturbed = np.clip(perturbed, min_, max_)
+
+            _, is_adversarial = a.predictions(perturbed)
+            if is_adversarial:
+                if abort_early or verify:
+                    break
+            else:
+                bound = epsilon
+                a.verifiable_bounds = (bound, None)
+
+        return