class SimpleCNN(torch.nn.Module):
def __init__(self, input_ch, output):
super(SimpleCNN, self).__init__()
self.cnn1 = torch.nn.Conv2d(in_channels=input_ch, out_channels=16, kernel_size=[3,3],stride=1, padding=1)
self.cnn2 = torch.nn.Conv2d(in_channels=16, out_channels=32, kernel_size=[3,3],stride=1, padding=1)
self.cnn3 = torch.nn.Conv2d(in_channels=32, out_channels=32, kernel_size=[3,3], padding=1)
self.cnn4 = torch.nn.Conv2d(in_channels=32, out_channels=64, kernel_size=[3,3], padding=1)
self.cnn5 = torch.nn.Conv2d(in_channels=64, out_channels=128, kernel_size=[3,3], padding=1)
self.cnn6 = torch.nn.Conv2d(in_channels=128, out_channels=256, kernel_size=[3,3], padding=1)
self.pool = torch.nn.MaxPool2d(kernel_size=[3,3])
self.Bnorm = torch.nn.BatchNorm2d(num_features=256)
self.drop = torch.nn.Dropout2d(p=0.3)
self.lin1 = torch.nn.Linear(256*7*7, out_features=512)
self.lin2 = torch.nn.Linear(512, output)
def forward(self, x):
x = F.relu(self.cnn1(x))
x = self.pool(F.relu(self.cnn2(x)))
x = F.relu(self.cnn3(x))
x = self.pool(F.relu(self.cnn4(x)))
x = F.relu(self.cnn5(x))
x = self.pool(F.relu(self.cnn6(x)))
x = self.Bnorm(x)
x = self.drop(x)
x = x.view(-1, 256*7*7)
x = F.relu(self.lin1(x))
x = self.lin2(x)
return x
model = SimpleCNN(3, 29)
My prediction (does this problem only for single image prediction)
#validation mode on torch, also there is a mode to turn the model into val() mode
correct = 0
total = 0
with torch.no_grad():
for data in test_dataLoader:
images, labels = data
outputs = net(images)
_, predicted = torch.max(outputs.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
print('Accuracy of the network on the 20 test images: %d %%' % (
100 * correct / total))