Hi guys,
I am trying to calculate the accuracy of my trained model(for multi-class image classification).
My labels are 1, 2, 3, 4.
My __getitem__ in data loader class looks like below -
def __getitem__(self, index):
"""Generate one sample of data."""
ID = self.list_IDs[index]
image = Image.open(os.path.join(self.dir, ID))
y = self.labels[ID]
if self.train_transform:
image = self.train_transforms(image)
else:
image = self.valid_transforms(image)
img = np.array(image)
return img, y
My test accuracy and loss function looks like below -
for batch_idx,(data,target) in enumerate(test_loader):
output = model_transfer(data).squeeze()
output = torch.unsqueeze(output,0)
loss = criterion(output, target)
avg_valid_loss += (1/(batch_idx+1)) * ((loss.data) - avg_valid_loss)
pred = output.data.max(1, keepdim=True)[1]
correct += np.sum(np.squeeze(pred.eq(target.data.view_as(pred))).cpu().numpy())
total += data.size(0)
accuracy = 100.*(correct/total)
Am I calculating the accuracy correctly?
Also, regarding the labels are 1, 2, 3, 4, should I convert them into some form of one-hot encoding or Is network able to predict it generally?
Any other suggestion regarding it is also welcome?