Jun-Yan Zhu

The code will preserve the L channel by default and only predict ab channels. If you really want to change the L channel, you can modify this [line](https://github.com/junyanz/interactive-deep-colorization/blob/master/ui/gui_draw.py#L282).

Not sure if this error is related to our program or Caffe installation. Have you run Caffe tests using`make runtest`?

You can add a flag `--cpu_mode`. It will work but it will be quite slow. If you just want to colorize your photo on Windows platform. you can use Adobe...

Maybe you want to double-check if the image `test_imgs/mortar_pestle.jpg` is in your directory. Maybe you need to rename the file according to Windows format.

I am not familiar with the Windows file path format. You should just go to python, and then debug the function `cv2.imread`. If you are not familiar with Python, you...

Not sure. If you only have one GPU, you should use gpu 0. You can also try cpu model with `--cpu_mode`.

Thank @SleepProgger for providing the installation guidelines. I updated the installation with more details. Let me know if I missed something.

Which color did you try to select? could you have a screenshot of UI?

I cannot reproduce your bug. The current system works for me. My python is python 2.7. I added a debugging [line](https://github.com/junyanz/interactive-deep-colorization/blob/master/ui/gui_palette.py#L73) that might help you find the issue.

Currently, only 256x models are available. One simple solution is that you upsample the chromatic prediction of the network to the original resolution, and combine it with the original grayscale...