Deep Learning with PyTorch

Classifying dogs and cats – CNN from scratch

We will use the same architecture with a few minor changes, as listed here:

• The input dimensions for the first linear layer changes, as the dimensions for our cats and dogs images are 256, 256
• We add another linear layer to give more flexibility for the model to learn

Let's look at the code that implements the network architecture:

class Net(nn.Module):    def __init__(self):        super().__init__()        self.conv1 = nn.Conv2d(3, 10, kernel_size=5)        self.conv2 = nn.Conv2d(10, 20, kernel_size=5)        self.conv2_drop = nn.Dropout2d()        self.fc1 = nn.Linear(56180, 500)        self.fc2 = nn.Linear(500,50)        self.fc3 = nn.Linear(50, 2)    def forward(self, x):        x = F.relu(F.max_pool2d(self.conv1(x), 2)) x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), ...