TypeError:Cannot handle this data type about ganimation HOT 11 OPEN

Not only covert it to tensor but convert to actual format (CXWXH) as follows:
I=torch.from_numpy(image_numpy).permute(2,0,1)
self._writer.add_image(sum_name, I/255, it)

from ganimation.

@Blade6570
I see, I replaced 'image_numpy' by 'np.transpose(image_numpy,(2,0,1))' and solved the problem.
Thank you very much.

from ganimation.

Seems this is a versions error. You are probably using Pytorch 0.4, this code was made previous to this version release. It was made for Pytorch 0.3.1.

from ganimation.

In this line:
File "/home/xirui/XIr/GANimation-master/utils/tb_visualizer.py", line 27, in display_current_results
self._writer.add_image(sum_name, image_numpy, it)self._writer.add_image(sum_name, image_numpy, it)
I changed the image_numpy in to tensor and normalized it between 0 to 1 which solved the problem

from ganimation.

@Blade6570
Hi, I also met the same problem. I changed the image_numpy into a tensor by using 'torch.from_numpy(image_numpy).div(255)' but still got stuck. Would you mind share your code with me?

Many thanks.

from ganimation.

@yanxirui Hi yanxirui, I met the same problem as you, may I know whether you have solved the problem?

from ganimation.

@yanxirui Hi yanxirui, I met the same problem as you, may I know whether you have solved the problem?

I find do_visuals may has no effect on the train itself so I made do_visuals = 0 before.... And I use the method @Blade6570 mentioned to solve this problem successfully. Thanks to him.

from ganimation.

@albertpumarola I have try pytorch 0.3.1,but it encounter a error in "display visualizer". Then I modified ganimation.py in Iine-346 def get_current_visuals(self): , and add "np.transpose(image_numpy,(2,0,1))" after numpy of OrderedDict()。
OR self._writer.add_image(sum_name, np.transpose(image_numpy,(2,0,1)), it) In "utils/tb_visualizer.py "...
I think it is caused by Different versions of tensorboardX ,not by the pytorch version。

from ganimation.

I think you should use tensorboard 1.8

from ganimation.

BTW, can you provide a docker file about the running environment? it's really helpful

from ganimation.

Use tensorboardX version-1.1. It looks like it works.

from ganimation.