CODEWITHSUNDEEP
pythonpytorchconv-neural-network
Burger
Burger

Reputation: 413

Visualizing CNN

Hi I'm trying to visualize CNN. I've been going through https://pytorch.org/tutorials/beginner/blitz/neural_networks_tutorial.html to study CNN by visualizing the structure. What I couldn't understand is its dimension.

enter image description here

import torch
import torch.nn as nn
import torch.nn.functional as F


class Net(nn.Module):

    def __init__(self):
        super(Net, self).__init__()
        # 1 input image channel, 6 output channels, 3x3 square convolution
        # kernel
        self.conv1 = nn.Conv2d(1, 6, 3)
        self.conv2 = nn.Conv2d(6, 16, 3)
        # an affine operation: y = Wx + b
        self.fc1 = nn.Linear(16 * 6 * 6, 120)  # 6*6 from image dimension
        self.fc2 = nn.Linear(120, 84)
        self.fc3 = nn.Linear(84, 10)

    def forward(self, x):
        # Max pooling over a (2, 2) window
        x = F.max_pool2d(F.relu(self.conv1(x)), (2, 2))
        # If the size is a square you can only specify a single number
        x = F.max_pool2d(F.relu(self.conv2(x)), 2)
        x = x.view(-1, self.num_flat_features(x))
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)
        return x

    def num_flat_features(self, x):
        size = x.size()[1:]  # all dimensions except the batch dimension
        num_features = 1
        for s in size:
            num_features *= s
        return num_features


net = Net()
print(net)

So the code part is supposed to be the CNN model structure from the image. What I don't get is this.

  1. Convolutions happen from the input to C1 and 3*3 kernel was used. In this case, shouldn't the dimension of C1 be 30 X 30 instead of 28 X 28?
  2. According to the image, dimension of the input to layer F5 is 16 X 5 X 5 but the code says otherwise. It seems like layer F5 is taking an input of dimension 16 X 6 X 6.

I'm not sure whether I'm taking it incorrectly or the image is wrong.

Upvotes: 0

Views: 228

Answers (1)

6etacat
6etacat

Reputation: 81

I'm pretty sure the image is wrong.

If you check the documentation of Conv2d. Using the equation there, the first convolution layer should output (batch_size, 6, 30, 30). Running the model also confirms my conclusion.

The image should be modified to:

INPUT: 1 x 32 x 32
C1: 6 x 30 x 30
S2: 6 x 15 x 15
C2: 16 x 13 x 13
S2: 16 x 6 x 6

Upvotes: 1

Related Questions