Update main.py

config.py

@@ -20,4 +20,5 @@
 imagenet10_valdir = 'C:/Users\Rui\Pictures/transfer_imgnet_10/val'
 
 models_path = './models/'
-adv_img_path = './images/0526/adv/'
+
+adv_img_path = './images/0526/adv/'

data/imagenet10_dataloader.py

@@ -12,7 +12,7 @@ def get_data_loaders():
     # Normalization transform: scales the image tensors to a standard range
     # These mean and std values are commonly used for ImageNet datasets
     normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406],
-                                     std=[0.229, 0.224, 0.225])  # Standard ImageNet normalization values
+                                     std=[0.229, 0.224, 0.225])  
 
     # Define the data loader for the training dataset
     train_loader = torch.utils.data.DataLoader(

deepsecure/adv_image.py

@@ -91,4 +91,5 @@ def train(self, train_dataloader, epochs):
                 netG_file_name = models_path + 'netG_epoch_' + str(epoch) + '.pth'
                 torch.save(self.generator.state_dict(), netG_file_name)
 
+
             print("check")

deepsecure/cat_adv_image.py

@@ -92,7 +92,7 @@
 #                 loss_img_sum += loss_img_batch
 
 
-#             # print statistics
+
 
 #             torchvision.utils.save_image(torch.cat((adv_img[:7], images[:7], (images[idx])[:7])),
 #                                          adv_img_path + str(epoch) + ".png",

deepsecure/catted_generator.py

@@ -162,7 +162,7 @@ def __init__(self, num_encoder_layers, fix_encoder, tagged):
                 nn.Tanh()
             ]
 
-        # Create sequential decoder module
+        # Create sequential decoder  module
         self.decoder = nn.Sequential(*decoder_lis)
 
     def forward(self, x1, x2):

deepsecure/regular_generator.py

@@ -77,5 +77,6 @@ def forward(self, x):
             x_t[:, :, :cfg.tag_size, :cfg.tag_size] = x_t.max()
         out = self.decoder(x_t)
 
+
         return out, x_t
 

main.py

@@ -37,9 +37,10 @@
     else:
         # Initialize the regular generator
          # Create adversarial generator with concatenated architecture
-        # False parameter indicates no regular generator for noise
+
         generator = regular_generator(cfg.num_layers_ext, cfg.ext_fixed, cfg.G_tagged)
         # Create an instance of Adv_Gen with the regular generator
+
         advGen = Adv_Gen(device, feature_ext, generator)