Thank you for your outstanding work in the area of image style transfer.

[shehuiyaoa11y]

I have been working with the AesPA-Net model for my image style transfer tasks and have been truly impressed by the significant contributions your work has made to this field. The results and the innovative approach you've taken are commendable.

However, I encountered an issue while testing the model. When running the following command:

python main.py --type test --batch_size 2 --comment 'Final_v9_adv_t2' --content_dir "/home/x/my_filer/AesPA-Net-main/test_images/content_final/" --style_dir "/home/x/my_filer/AesPA-Net-main/test_images/style_final/" --num_workers 1

I received a warning related to an empty slice and invalid values in numpy:

/home/x/anaconda3/envs/AesPA/lib/python3.7/site-packages/numpy/core/fromnumeric.py:3441: RuntimeWarning: Mean of empty slice.
/home/x/anaconda3/envs/AesPA/lib/python3.7/site-packages/numpy/core/_methods.py:189: RuntimeWarning: invalid value encountered in double_scalars

The final output shows avg: nan. I have reviewed the inputs and settings but couldn't identify the root cause of the issue. Could you kindly provide guidance on how to resolve this?

I greatly appreciate your time and assistance. Your insights would be invaluable as I work towards better understanding and implementing your outstanding model.

Thank you once again for your exceptional work in the field of image style transfer.

[Kibeom-Hong]

I appreciate your interest of my research. I think the input value(maybe images or calculated intermediate values) contain the "NaN" value while calculated by np.mean().

So I recommend that you could check which line occurs that error, and I think you should replace np.mean() with np.nanmean()

[shehuiyaoa11y]

I'll try that. Thanks again.

[ciyuanjing]

I'll try that. Thanks again.

I think your path designation is wrong, not the mean problem, because I just had the same error because the path designation is wrong