Error on single image inference, thanks

[junyongyou]

I have tried to use the model to do aesthetics assessment on a single image, and below is my code snippet:

`
class IAA:
def init(self, model_file, weights_path):
self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
self.init_model(model_file, weights_path)
normalize = transforms.Normalize(mean=IMAGE_NET_MEAN, std=IMAGE_NET_STD)
self.transform = transforms.Compose([
transforms.ToTensor(),
normalize
])

def init_model(self, model_file, weights_path):
    self.model = TANet(model_file)
    self.model.load_state_dict(torch.load(weights_path))
    self.model.to(self.device)
    self.model.eval()

def inference(self, image_path):
    with open(image_path, 'rb') as f:
        image = Image.open(f)
        image = image.convert('RGB')
        image = image.resize((224, 224))
        image = self.transform(image)
        image = torch.unsqueeze(image, dim=0)
        image = image.to(self.device)
        iia_score = self.model(image)
    return iia_score

if name == 'main':
weights_path = r'.\weights\SRCC_758_LCC_765.pth'
model_file = r'.\weights\resnet18_places365.pth.tar'
image_path = r'.\samples\00000.png'
iaa = IAA(model_file, weights_path)
score = iaa.inference(image_path)

`

However, I got an error: ValueError: Expected more than 1 value per channel when training, got input size torch.Size([1, 1]).

It seems this is a reported error when using batchnorm on a single image (batch size =1), see here (https://discuss.pytorch.org/t/error-expected-more-than-1-value-per-channel-when-training/26274/67). However, it was suggested that adding model.eval() can solve the problem. As you can see, I used model.eval(), but still got the error.

Did you happen to encounter to this error? If so, how did you solve it? Thanks a lot.

[junyongyou]

My current workaround solution is to use multiple images in a batch. However, it might be better to figure out how to solve the single image issue.

In addition, the model produces a softmax output with 10 element for a single image, because I am using the AVA model. How can we compute a single score for an image? Thank you.

[woshidandan]

1. I have never encountered such a problem before ，you may refine the function of "validate(opt, model=model, loader=dataloader_test……" when bs=1，to see if you can solve this problem，see issue 8.
2. use the function of "def get_score(opt, y_pred)" in my code to obtain a score.

[junyongyou]

1. I have never encountered such a problem before ，you may refine the function of "validate(opt, model=model, loader=dataloader_test……" when bs=1，to see if you can solve this problem，see issue 8.

   2. use the function of "def get_score(opt, y_pred)" in my code to obtain a score.

I got it, thanks a lot.

[justlike-prog]

@junyongyou hi could you solve this issue without using the dataloader? For now I am doing

bn7 = nn.BatchNorm1d(q.shape[0])
bn7.eval()

because it seems that the problem is that the BatchNorm1d is in training mode, although we set the whole model to eval. I am sure there is nicer solution. I guess the BatchNorm should be created when creating the TargetNet and then the forward pass just calls it.

[justlike-prog]

@woshidandan Ok my solution doesn't work because it throws some wrong predictions. I would need the single image inference without dataloader for converting the model to onnx. Any ideas?

[junyongyou]

There are some sub-models that are not in eval mode even the whole model is set to eval. Thus, I don't know how to solve the problem.

[chenminupup]

@junyongyou 关于single image inference，想问下你这边解决了吗？