Jun-Yan Zhu

It seems that opencv doesn't work on your machine. You could print the image_file. You can also write a simple test script to test the OpenCV. ```python import cv2 im_bgr...

hmmm... maybe removing the `.encode('utf-8')`. also print out the image_file in the gui_draw.py,

Haven't seen that before. Maybe you would like to print out the color_id variable.

Thanks for the update. I fixed the integer division issue with the latest commit.

probably. It was introduced by this [PR](https://github.com/junyanz/interactive-deep-colorization/pull/23).

@mrlzc Have you tried to compile the Caffe with the python layer support? See this [post](https://chrischoy.github.io/research/caffe-python-layer/) for more details.

I can run the program on a GTX 1080 with 8 GB memory. The program consumes 4.5 GB GPU memory for the default test image `mortar_pestle.jpg`.

Sorry to hear that. The current code has not been tested on OSX (as we don't have a MAC with fancy GPUs. )

Thanks a lot! @vbisbest

Are you able to run the basic Caffe test code?