Prediction on Image and mask without UI

[kashyappiyush1998]

I have a binary mask and image to be filled and want to predict output, how do I do that using this repository. Like in deep fill v1/v2 we can predict it directly. Thank you. Results are pretty awesome on comodgan.ml

[styler00dollar]

I agree. I would be interested to use co-mod-gan inside Google Colab and that needs command line. I tried to modify run_demo.py, but I got many errors.

[zsyzzsoft]

Now one can create a custom dataset with dataset_tools/create_from_images and then run demo with it!

[kashyappiyush1998]

@zsyzzsoft thanks for this added tool, but is it not possible to directly get the predicted image out of the image and mask without running UI. If possible, I have one more request, can you provide a sample of inputs both image and mask along with it.

Thanks.

[zsyzzsoft]

@zsyzzsoft thanks for this added tool, but is it not possible to directly get the predicted image out of the image and mask without running UI. If possible, I have one more request, can you provide a sample of inputs both image and mask along with it.

Thanks.

Please try the newly added run_generator.py.

[styler00dollar]

Thanks, it works. Issue can be closed.

[tiwarikaran]

@styler00dollar hi can you share the notebook if possible I am having a tough time implementing this

[styler00dollar]

@tiwarikaran Well I am pretty late, but my colab notebook is here.