CODEWITHSUNDEEP
pythonmachine-learningconv-neural-networkpytorchautoencoder
Talal Zahid
Talal Zahid

Reputation: 128

Convolution Autoencoder Image Dimension Error

I have the following Convolutional Autoencoder setup:

class autoencoder(nn.Module):
def  __init__(self):
    super(autoencoder, self).__init__()
    self.encoder = nn.Sequential(
        nn.Conv2d(1, 16, 3, stride=3, padding=1),  # b, 16, 10, 10
        nn.ReLU(True),
        nn.MaxPool2d(2, stride=2),  # b, 16, 5, 5
        nn.Conv2d(16, 8, 3, stride=2, padding=1),  # b, 8, 3, 3
        nn.ReLU(True),
        nn.MaxPool2d(2, stride=1)  # b, 8, 2, 2                     
                                )

    self.decoder = nn.Sequential(
        nn.ConvTranspose2d(8, 16, 3, stride=2),  # b, 16, 5, 5
        nn.ReLU(True),
        nn.ConvTranspose2d(16, 8, 5, stride=3, padding=1),  # b, 8, 15, 15
        nn.ReLU(True),
        nn.ConvTranspose2d(8, 1, 2, stride=2, padding=1),  # b, 1, 28, 28
        nn.Tanh()          
                                )

This is the main Loop:

for epoch in range(epochs):
running_loss = 0
for data in (train_loader):
    image,_=data
    inputs = image.view(image.size(0),-1)
    optimizer.zero_grad()
    #image = np.expand_dims(img, axis=0)

    outputs = net(inputs)
    loss = criterion(outputs,inputs)
    loss.backward()
    optimizer.step()
    running_loss += loss.data[0]
print('At Iteration : %d   ;  Mean-Squared Error : %f'%(epoch + 1,running_loss/(train_set.train_data.size(0)/batch_size)))

This is the Error:

RuntimeError: Expected 4-dimensional input for 4-dimensional weight [16, 1, 3, 3], but got input of size [1000, 784] instead

This has something to do with the flattening of the image but Im not exactly sure how to deflatten it.

Upvotes: 0

Views: 165

Answers (1)

Shai
Shai

Reputation: 114796

Why are you "flattening" your input image (2nd line of main loop):

inputs = image.view(image.size(0),-1)

This line turns your 4 dimensional image (batch - channels - height - width) to a two dimensional "flat" vector (batch - c * h * w).
You autoencoder expects its inputs to be 4D and not "flat". just remove this line and you should be okay.

Upvotes: 1

Related Questions