DenseNet Error

[Pen721]

Hi! I'm working through the tutorial and trying to make a visualization for a 96x96x3 image using DenseNet but running into the following error, wondering if there's anyway I can fix this.

Thanks so much!

model = model1.model

model.eval()
model.features.denseblock4.denselayer16.eval()
# image_url = ''
img = img #np.array(Image.open(requests.get(image_url, stream=True).raw))
# img = cv2.resize(img, (224, 224))
# img = np.float32(img) / 255
# input_tensor = preprocess_image(img, mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])

# The target for the CAM is the Bear category.
# As usual for classication, the target is the logit output
# before softmax, for that category.
targets = [BinaryClassifierOutputTarget(1)]
target_layers = [model.features.denseblock4.denselayer16.conv1, model.features.denseblock4.denselayer16.norm2]

with GradCAM(model=model, target_layers=target_layers) as cam:
    grayscale_cams = cam(input_tensor=img, targets=targets)
    cam_image = show_cam_on_image(img, grayscale_cams[0, :], use_rgb=True)
cam = np.uint8(255*grayscale_cams[0, :])
cam = cv2.merge([cam, cam, cam])
images = np.hstack((np.uint8(255*img), cam , cam_image))
Image.fromarray(images)

---

RuntimeError                              Traceback (most recent call last)
Input In [67], in <cell line: 17>()
     15 target_layers = [model.features.denseblock4.denselayer16.conv1, model.features.denseblock4.denselayer16.norm2]
     17 with GradCAM(model=model, target_layers=target_layers) as cam:
---> 18     grayscale_cams = cam(input_tensor=img, targets=targets)
     19     cam_image = show_cam_on_image(img, grayscale_cams[0, :], use_rgb=True)
     20 cam = np.uint8(255*grayscale_cams[0, :])

File path/python3.8/site-packages/pytorch_grad_cam/base_cam.py:188, in BaseCAM.__call__(self, input_tensor, targets, aug_smooth, eigen_smooth)
    184 if aug_smooth is True:
    185     return self.forward_augmentation_smoothing(
    186         input_tensor, targets, eigen_smooth)
--> 188 return self.forward(input_tensor,
    189                     targets, eigen_smooth)

File path/site-packages/pytorch_grad_cam/base_cam.py:74, in BaseCAM.forward(self, input_tensor, targets, eigen_smooth)
     70 if self.compute_input_gradient:
     71     input_tensor = torch.autograd.Variable(input_tensor,
     72                                            requires_grad=True)
---> 74 outputs = self.activations_and_grads(input_tensor)
     75 if targets is None:
     76     target_categories = np.argmax(outputs.cpu().data.numpy(), axis=-1)

File path/site-packages/pytorch_grad_cam/activations_and_gradients.py:42, in ActivationsAndGradients.__call__(self, x)
     40 self.gradients = []
     41 self.activations = []
---> 42 return self.model(x)

File /path/site-packages/torch/nn/modules/module.py:727, in Module._call_impl(self, *input, **kwargs)
    725     result = self._slow_forward(*input, **kwargs)
    726 else:
--> 727     result = self.forward(*input, **kwargs)
    728 for hook in itertools.chain(
    729         _global_forward_hooks.values(),
    730         self._forward_hooks.values()):
    731     hook_result = hook(self, input, result)

File path/site-packages/torchvision/models/densenet.py:192, in DenseNet.forward(self, x)
    191 def forward(self, x):
--> 192     features = self.features(x)
    193     out = F.relu(features, inplace=True)
    194     out = F.adaptive_avg_pool2d(out, (1, 1))

File path/site-packages/torch/nn/modules/module.py:727, in Module._call_impl(self, *input, **kwargs)
    725     result = self._slow_forward(*input, **kwargs)
    726 else:
--> 727     result = self.forward(*input, **kwargs)
    728 for hook in itertools.chain(
    729         _global_forward_hooks.values(),
    730         self._forward_hooks.values()):
    731     hook_result = hook(self, input, result)

File path/site-packages/torch/nn/modules/container.py:117, in Sequential.forward(self, input)
    115 def forward(self, input):
    116     for module in self:
--> 117         input = module(input)
    118     return input

File path/site-packages/torch/nn/modules/module.py:727, in Module._call_impl(self, *input, **kwargs)
    725     result = self._slow_forward(*input, **kwargs)
    726 else:
--> 727     result = self.forward(*input, **kwargs)
    728 for hook in itertools.chain(
    729         _global_forward_hooks.values(),
    730         self._forward_hooks.values()):
    731     hook_result = hook(self, input, result)

File path/site-packages/torch/nn/modules/conv.py:423, in Conv2d.forward(self, input)
    422 def forward(self, input: Tensor) -> Tensor:
--> 423     return self._conv_forward(input, self.weight)

File path/site-packages/torch/nn/modules/conv.py:419, in Conv2d._conv_forward(self, input, weight)
    415 if self.padding_mode != 'zeros':
    416     return F.conv2d(F.pad(input, self._reversed_padding_repeated_twice, mode=self.padding_mode),
    417                     weight, self.bias, self.stride,
    418                     _pair(0), self.dilation, self.groups)
--> 419 return F.conv2d(input, weight, self.bias, self.stride,
    420                 self.padding, self.dilation, self.groups)

**RuntimeError: Expected 4-dimensional input for 4-dimensional weight [64, 3, 7, 7], but got 3-dimensional input of size [96, 96, 3] instead**

[jacobgil]

Hi,
The image that's sent to the model has to be a tensor, pre-processed (if any preprocessing is required, like for pretrained models on imagenet), and with shape batch x 3 x height x width

In your case the tensor has shape 96, 96, 3
You can make it 1 x 3 x 96 x 96 like this:
img = img.transpose(2, 0, 1).unsqueeze(0

But let me ask you to be sure:

• Is img a tensor ?
• Did you already take care of image-net normalization ?

[Pen721]

Thank you so much! This was super helpful, that was exactly it