search

TachibanaYoshino / AnimeGAN

A Tensorflow implementation of AnimeGAN for fast photo animation ! This is the Open source of the paper 「AnimeGAN: a novel lightweight GAN for photo animation」, which uses the GAN framwork to transform real-world photos into anime images.
4.49k stars 660 forks source link

Installation scripts and instructions in the README #13

Closed iacolippo closed 1 year ago

iacolippo commented 4 years ago

Hi, I love your work with AnimeGAN. I took some time to try it out and it's really fun.

Since I encountered some hiccups in the installation, I have written some scripts and a detailed update to the README to install the right versions of CUDA, cuDNN and Tensorflow. I also added the instructions on how to write AnimeGAN on user-chosen images.

Hope this is helpful, cheers

TachibanaYoshino commented 4 years ago

Thank you very much. For installing cuda and cudnn, my suggestion is to use anaconda or miniconda to create a new independent python environment. This new environment is independent of the system python. In the new environment, you can easily and quickly build GPU support by using the commands conda install cuda = 9.0 and conda install cudnn = 7.1.3. Then use the command pip install tensorflow-gpu == 1.8.0, pip install numpy, etc. to install other Dependency package. The independent environment built in this way can support the current project operation and development. When other projects need different cuda and cudnn versions in the future, you can also create a new environment through conda and install new cuda and cudnn in the environment. In this way, you can avoid installing tools such as cuda directly based on the system environment. By parsing the code of train.py and test.py, you should know how to perform training and testing correctly.

These are some personal thoughts. Thank you sincerely.

iacolippo commented 4 years ago

Oh wow, I didn't know you could do that with conda, then yes, that's a much easier and portable approach. Maybe it's worth it to add it in the README for clueless people like me :-)