Legacy autograd function with non-static forward method is deprecated. Please use new-style autograd function with static forward method. (Example: https://pytorch.org/docs/stable/autograd.html#torch.autograd.Function)

[ES95]

• 760 - 07-May-20 19:39:01 - Finished loading model!
• 760 - 07-May-20 19:39:02 - Test dataset size is 3162
• 760 - 07-May-20 19:39:03 - processing 8/3162
• 760 - 07-May-20 19:39:06 - Exception occurred
• Traceback (most recent call last):
• File "test.py", line 196, in

• test_gtdb(args)
  

• File "test.py", line 128, in test_gtdb

• thresh=args.visual_threshold)
  

• File "test.py", line 54, in test_net_batch

• y, debug_boxes, debug_scores = net(images)  # forward pass
  

• File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 550, in call

• result = self.forward(*input, **kwargs)
  

• File "/usr/local/lib/python3.6/dist-packages/torch/nn/parallel/data_parallel.py", line 153, in forward

• return self.module(*inputs[0], **kwargs[0])
  

• File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 550, in call

• result = self.forward(*input, **kwargs)
  

• File "/content/gdrive/My Drive/ssd/ScanSSD/ssd.py", line 112, in forward

• self.priors.type(type(x.data))                  # default boxes
  

• File "/usr/local/lib/python3.6/dist-packages/torch/autograd/function.py", line 145, in call

• _"Legacy autograd function with non-static forward method is deprecated. "
  

• RuntimeError: Legacy autograd function with non-static forward method is deprecated. Please use new-style autograd function with static forward method. (Example: https://pytorch.org/docs/stable/autograd.html#torch.autograd.Function)_
• 760 - 07-May-20 19:39:06 - Toal time taken 0:00:08.179874
• 760 - 07-May-20 19:39:06 - Testing done!

and i am just using two pages from pdf in gtdb dataset, is this correct "Test dataset size is 3162"

[ES95]

and i am just using two pages from pdf in gtdb dataset, is this correct "Test dataset size is 3162"

[MaliParag]

Yes. We generate page image crops using sliding windows.

[ES95]

And what about the compatibility issue RuntimeError: Legacy autograd function with non-static forward method is deprecated. Please use new-style autograd function with static forward method

[MaliParag]

Are you using the pytorch and cuda version mentioned in the readme introduction?

[ES95]

nope, i was running this code in colab and they don't have an option to downgrade/uninstall CUDA drivers. if you know a fix , do let me know

[jeannefukumaru]

This Runtime Error is something implemented in more recent versions of Pytorch. See this Pytorch discussion forum thread for more context.

Basically the layers/detection.py needs a @staticmethod decorator on its forward method. see this GitHub repo for making the necessary updates to the code.

happy to make a merge request here if needed

[siriasadeddin]

Use python 3.6 and an older version of pytorch. conda install pytorch==1.2.0 torchvision==0.4.0 -c pytorch