import numpy as np
import matplotlib.pyplot as plt
from utils import display_image, mean_cov
from attack import PGD_attack_fast
#loading the training dataset
train_cat = np.matrix(np.loadtxt('./dataset/train_cat.txt', delimiter=','))
train_grass = np.matrix(np.loadtxt('./dataset/train_grass.txt', delimiter=','))
#loading the test dataset
Y = plt.imread('./dataset/cat_grass.jpg') / 255
truth = plt.imread('./dataset/truth.png')
#computing the parameters for gaussian classifier (Training)
mean_cat, cov_cat, pi_cat = mean_cov(train_cat, train_grass)
mean_grass, cov_grass, pi_grass = mean_cov(train_grass, train_cat)
#overlapping
stride = 1
#Inference
display_image(img_perturbed=Y,
mean_cat=mean_cat,
cov_cat=cov_cat,
pi_cat=pi_cat,
mean_grass=mean_grass,
cov_grass=cov_grass,
pi_grass=pi_grass,
original_img=Y,
truth=truth,
Adv_training_data = Adv_training_data_CW
elif args.train=='PGD':
Adv_training_data = Adv_training_data_PGD
else:
raise ValueError
#loading the training dataset
train_cat=np.matrix(np.loadtxt('./dataset/train_cat.txt',delimiter=','))
train_grass=np.matrix(np.loadtxt('./dataset/train_grass.txt',delimiter=','))
#loading the test dataset
Y = plt.imread ('./dataset/cat_grass.jpg')/255
truth = plt.imread ('./dataset/truth.png')
#computing the parameters for gaussian classifier (Training)
mean_cat,cov_cat, pi_cat = mean_cov(train_cat,train_grass)
mean_grass,cov_grass, pi_grass = mean_cov(train_grass,train_cat)
Lamda = np.linspace (0,5.5,10)
Alpha = np.linspace (0.00002, 0.0003,10)
augmented_cat =train_cat
augmented_grass = train_grass
#parameters for blackbox attack
mean_cat_defense = mean_cat
cov_cat_defense = cov_cat
pi_cat_defense = pi_cat
mean_grass_defense = mean_grass
cov_grass_defense = cov_grass
pi_grass_defense = pi_grass
if args.train == 'CW':
Adv_training_data = Adv_training_data_CW
elif args.train == 'PGD':
Adv_training_data = Adv_training_data_PGD
else:
raise ValueError
#loading the training dataset
train_cat = np.matrix(np.loadtxt('./dataset/train_cat.txt', delimiter=','))
train_grass = np.matrix(np.loadtxt('./dataset/train_grass.txt', delimiter=','))
#loading the test dataset
Y = plt.imread('./dataset/cat_grass.jpg') / 255
truth = plt.imread('./dataset/truth.png')
#computing the parameters for gaussian classifier (Training)
mean_cat, cov_cat, pi_cat = mean_cov(train_cat, train_grass)
mean_grass, cov_grass, pi_grass = mean_cov(train_grass, train_cat)
Lamda = np.linspace(0.5, 5.5, 10)
Alpha = np.linspace(0.00002, 0.0003, 10)
augmented_cat = train_cat
augmented_grass = train_grass
#parameters for blackbox attack
mean_cat_defense = mean_cat
cov_cat_defense = cov_cat
pi_cat_defense = pi_cat
mean_grass_defense = mean_grass
cov_grass_defense = cov_grass
pi_grass_defense = pi_grass
global args
args = parser.parse_args()
print(args)
#loading the training dataset
train_cat=np.matrix(np.loadtxt('./dataset/train_cat.txt',delimiter=','))
train_grass=np.matrix(np.loadtxt('./dataset/train_grass.txt',delimiter=','))
#loading the test dataset
Y = plt.imread ('./dataset/cat_grass.jpg')/255
truth = plt.imread ('./dataset/truth.png')
#computing the parameters for gaussian classifier (Training)
q1 = Quantize(n_bits=args.n_bits, quantize=args.quantize)
mean_cat_defense,cov_cat_defense, pi_cat_defense = mean_cov(q1.forward(train_cat),q1.forward(train_grass))
mean_grass_defense,cov_grass_defense, pi_grass_defense = mean_cov(q1.forward(train_grass),q1.forward(train_cat))
if args.attack_type.lower() == 'whitebox':
mean_cat_attack = mean_cat_defense
cov_cat_attack = cov_cat_defense
pi_cat_attack = pi_cat_defense
mean_grass_attack = mean_grass_defense
cov_grass_attack = cov_grass_defense
pi_grass_attack = pi_grass_defense
elif args.attack_type.lower() == 'blackbox':
mean_cat_attack,cov_cat_attack, pi_cat_attack = mean_cov(train_cat,train_grass)
mean_grass_attack,cov_grass_attack, pi_grass_attack = mean_cov(train_grass,train_cat)
print("Inference without Quantization")
#Inference
display_image(img_perturbed = Y,