from perceptron.utils.criteria.classification import TopKMisclassification
from perceptron.utils.tools import plot_image
from perceptron.utils.tools import bcolors
# instantiate the model from keras applications
xception = models.Xception(weights='imagenet')
# initialize the KerasModel
# keras xception has input bound (0, 1)
mean = np.array([0.485, 0.456, 0.406]).reshape((1, 1, 3))
std = np.array([0.229, 0.224, 0.225]).reshape((1, 1, 3))
kmodel = KerasModel(xception, bounds=(0, 1), preprocessing=(mean, std))
# get source image and label
# the model Xception expects values in [0, 1] with shape (299, 299), and channles_last
image, _ = imagenet_example(shape=(299, 299), data_format='channels_last')
image /= 255.0
label = np.argmax(kmodel.predictions(image))
metric = AdditiveGaussianNoiseMetric(kmodel,
criterion=TopKMisclassification(10))
print(bcolors.BOLD + 'Process start' + bcolors.ENDC)
adversary = metric(
image, label, unpack=False,
epsilons=1000) # choose 1000 different epsilon values in [0, 1]
print(bcolors.BOLD + 'Process finished' + bcolors.ENDC)
if adversary.image is None:
print(bcolors.WARNING + 'Warning: Cannot find an adversary!' +
bcolors.ENDC)
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(