Upload neural_style.py

neural_style.py

@@ -59,7 +59,7 @@ class TransferParams():
     style_blend_weights = None
     content_image = 'examples/inputs/tubingen.jpg'
     image_size = 512
-    gpu = 0
+    gpu = "c"
     content_weight = 5e0
     style_weight = 1e2
     normalize_weights = False
@@ -71,7 +71,7 @@ class TransferParams():
     learning_rate = 1e0
     lbfgs_num_correction = 100
     print_iter = 50
-    save_iter = 100
+    save_iter = 1000
     output_image = 'out.png'
     log_level = 10
     style_scale = 1.0
@@ -79,7 +79,7 @@ class TransferParams():
     pooling = 'max'
     model_file = 'models/vgg19-d01eb7cb.pth'
     disable_check = False
-    backend = 'nn'
+    backend = 'mkl'
     cudnn_autotune = False
     seed = -1
     content_layers = 'relu4_2'
@@ -91,6 +91,7 @@ def main():
 
 def transfer(params):
     dtype, multidevice, backward_device = setup_gpu()
+    print(dtype)
 
     cnn, layerList = loadCaffemodel(params.model_file, params.pooling, params.gpu, params.disable_check)
 
@@ -238,7 +239,7 @@ def transfer(params):
         else:
             img = content_image.clone()
     img = nn.Parameter(img)
-
+    
     def maybe_print(t, loss):
         if params.print_iter > 0 and t % params.print_iter == 0:
             print("Iteration " + str(t) + " / "+ str(params.num_iterations))
@@ -248,8 +249,9 @@ def transfer(params):
                 print("  Style " + str(i+1) + " loss: " + str(loss_module.loss.item()))
             print("  Total loss: " + str(loss.item()))
 
+    final_image = ''
     def maybe_save(t):
-        should_save = params.save_iter > 0 and t % params.save_iter == 0
+        should_save = params.save_iter > 950 and t % params.save_iter == 0
         should_save = should_save or t == params.num_iterations
         if should_save:
             output_filename, file_extension = os.path.splitext(params.output_image)
@@ -263,7 +265,11 @@ def transfer(params):
             if params.original_colors == 1:
                 disp = original_colors(deprocess(content_image.clone()), disp)
 
+            print('Saving')
             disp.save(str(filename))
+            print('the final image is:', disp)
+            final_image = disp
+            return disp
 
     # Function to evaluate loss and gradient. We run the net forward and
     # backward to get the gradient, and sum up losses from the loss modules.
@@ -271,6 +277,7 @@ def transfer(params):
     # times, so we manually count the number of iterations to handle printing
     # and saving intermediate results.
     num_calls = [0]
+
     def feval():
         num_calls[0] += 1
         optimizer.zero_grad()
@@ -287,16 +294,16 @@ def transfer(params):
 
         loss.backward()
 
-        maybe_save(num_calls[0])
+        final_image = maybe_save(num_calls[0])
         maybe_print(num_calls[0], loss)
 
         return loss
-
+    print('the final image is', final_image)
     optimizer, loopVal = setup_optimizer(img)
     while num_calls[0] <= loopVal:
          optimizer.step(feval)
 
-
+    
 # Configure the optimizer
 def setup_optimizer(img):
     if params.optimizer == 'lbfgs':
@@ -347,9 +354,9 @@ def setup_gpu():
             setup_cuda()
         dtype = torch.FloatTensor
 
-    elif "c" not in str(params.gpu).lower():
-        setup_cuda()
-        dtype, backward_device = torch.cuda.FloatTensor, "cuda:" + str(params.gpu)
+    #elif "c" not in str(params.gpu).lower():
+        #setup_cuda()
+        #dtype, backward_device = torch.cuda.FloatTensor, "cuda:" + str(params.gpu)
     else:
         setup_cpu()
         dtype, backward_device = torch.FloatTensor, "cpu"