def __init__(self, models, data, class_names, dimensions=(32, 32)):
# Load data and model
self.models = models
self.x_test, self.y_test = data
self.class_names = class_names
self.dimensions = dimensions
network_stats, correct_imgs = helper.evaluate_models(self.models, self.x_test, self.y_test)
self.correct_imgs = pd.DataFrame(correct_imgs, columns=['name', 'img', 'label', 'confidence', 'pred'])
self.network_stats = pd.DataFrame(network_stats, columns=['name', 'accuracy', 'param_count'])
def __init__(self, models, data, class_names, dimensions=(28, 28)):
# Load data and model
self.models = models
self.x_test, self.y_test = data
self.class_names = class_names
self.dimensions = dimensions
network_stats, correct_imgs = helper.evaluate_models(self.models, self.x_test, self.y_test)
self.correct_imgs = pd.DataFrame(correct_imgs, columns=['name', 'img', 'label', 'confidence', 'pred'])
self.network_stats = pd.DataFrame(network_stats, columns=['name', 'accuracy', 'param_count'])
# processed_images = mobilenet_preprocess_input(images.copy())
processed_images = images.copy()
print("Processed Images shape {}, min {}, max {}".format(
processed_images.shape, np.min(processed_images),
np.max(processed_images)))
# # -----------------------------------------------------------------------------------
# # Model Accuracy
# # -----------------------------------------------------------------------------------
print("Evaluating Model Accuracy - Regular Images")
models = [model]
network_stats, correct_imgs = helper.evaluate_models(
models,
processed_images,
labels,
preprocessing_cb=mobilenet_preprocess_input)
correct_imgs = pd.DataFrame(
correct_imgs, columns=['name', 'img', 'label', 'confidence', 'pred'])
network_stats = pd.DataFrame(network_stats,
columns=['name', 'accuracy', 'param_count'])
print(network_stats)
print("Number of correctly Classifiered images {}".format(
len(correct_imgs)))
# -----------------------------------------------------------------------------------
# Adverserial Attacks
# -----------------------------------------------------------------------------------
batchsize=10)
images = np.array([img_to_array(original) for original in images
]) # Convert the PIL image to a numpy array
#print(np.shape(y_test))
print(np.shape(labels.reshape(10, 1)))
labels = labels.reshape(10, 1)
#自定义网络
# lenet = LeNet()
resnet = ResNet()
# model_filename = 'networks/models/resnet.h5'
# resnet=load_model(model_filename)
models = [resnet]
network_stats, correct_imgs = helper.evaluate_models(models, images, labels)
correct_imgs = pd.DataFrame(
correct_imgs, columns=['name', 'img', 'label', 'confidence', 'pred'])
network_stats = pd.DataFrame(network_stats,
columns=['name', 'accuracy', 'param_count'])
print(network_stats)
pixel = np.array([16, 20, 0, 255, 255])
model = resnet
image_id = 1
true_class = labels[image_id, 0]
prior_confidence = model.predict_one(images[image_id])[true_class]
confidence = helper.predict_classes(pixel, images[image_id], true_class,
model)[0]
print('Attacking with target', class_names[target])
if (target == y_test[img, 0]):
continue
result = attack(img,
model,
target,
pixel_count,
maxiter=args.maxiter,
popsize=args.popsize,
verbose=verbose)
model_results.append(result)
results += model_results
return results
network_stats, correct_imgs = evaluate_models(models, x_test, y_test)
correct_imgs = pd.DataFrame(
correct_imgs, columns=['name', 'img', 'label', 'confidence', 'pred'])
network_stats = pd.DataFrame(network_stats,
columns=['name', 'accuracy', 'param_count'])
print('Starting attack')
results = attack_all(models,
samples=args.samples,
targeted=args.targeted,
pixels=args.pixels,
verbose=args.verbose)
print('Saving to', args.savedir)
with open(args.savedir, 'wb') as file:
