Update dataloder_pytorch.py
Browse files- dataloder_pytorch.py +60 -0
dataloder_pytorch.py
CHANGED
@@ -46,3 +46,63 @@ val_data = DataLoader(CarShadowDataset(root_dir='dataset/val', transform=your_tr
|
|
46 |
for car_image, shadow_image in train_data:
|
47 |
# Access your data for training
|
48 |
# ...
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
for car_image, shadow_image in train_data:
|
47 |
# Access your data for training
|
48 |
# ...
|
49 |
+
|
50 |
+
|
51 |
+
# ... (Previous code for model definition and DataLoader)
|
52 |
+
|
53 |
+
# Discriminator Training
|
54 |
+
def train_discriminator(d_optimizer, real_images, fake_images, real_labels, fake_labels):
|
55 |
+
# Clear gradients
|
56 |
+
d_optimizer.zero_grad()
|
57 |
+
|
58 |
+
# Forward pass through discriminator
|
59 |
+
d_real_output = discriminator(real_images, real_images) # Real images with real shadows
|
60 |
+
d_fake_output = discriminator(real_images, fake_images) # Real images with generated shadows
|
61 |
+
|
62 |
+
# Calculate loss
|
63 |
+
d_real_loss = criterion(d_real_output, torch.ones_like(d_real_output))
|
64 |
+
d_fake_loss = criterion(d_fake_output, torch.zeros_like(d_fake_output))
|
65 |
+
d_loss = (d_real_loss + d_fake_loss) / 2
|
66 |
+
|
67 |
+
# Backpropagate and update weights
|
68 |
+
d_loss.backward()
|
69 |
+
d_optimizer.step()
|
70 |
+
|
71 |
+
# Return the discriminator loss
|
72 |
+
return d_loss.item()
|
73 |
+
|
74 |
+
# Generator Training
|
75 |
+
def train_generator(g_optimizer, real_images, fake_images):
|
76 |
+
# Clear gradients
|
77 |
+
g_optimizer.zero_grad()
|
78 |
+
|
79 |
+
# Forward pass through discriminator (using generated shadows)
|
80 |
+
g_fake_output = discriminator(real_images, fake_images)
|
81 |
+
|
82 |
+
# Calculate loss (try to fool the discriminator)
|
83 |
+
g_loss = criterion(g_fake_output, torch.ones_like(g_fake_output))
|
84 |
+
|
85 |
+
# Backpropagate and update weights
|
86 |
+
g_loss.backward()
|
87 |
+
g_optimizer.step()
|
88 |
+
|
89 |
+
# Return the generator loss
|
90 |
+
return g_loss.item()
|
91 |
+
|
92 |
+
# Training loop
|
93 |
+
for epoch in range(epochs):
|
94 |
+
for i, (real_images, real_shadows) in enumerate(train_data):
|
95 |
+
# Generate fake shadows
|
96 |
+
fake_shadows = generator(real_images)
|
97 |
+
|
98 |
+
# Train discriminator
|
99 |
+
d_loss = train_discriminator(d_optimizer, real_images, fake_shadows, torch.ones(real_images.size(0)), torch.zeros(real_images.size(0)))
|
100 |
+
|
101 |
+
# Train generator
|
102 |
+
g_loss = train_generator(g_optimizer, real_images, fake_shadows)
|
103 |
+
|
104 |
+
# Print training progress
|
105 |
+
if i % 100 == 0:
|
106 |
+
print(f'Epoch [{epoch+1}/{epochs}], Step [{i+1}/{len(train_data)}], D_loss: {d_loss:.4f}, G_loss: {g_loss:.4f}')
|
107 |
+
|
108 |
+
|