Documentation • Contributing • Authors • License
Made by Bilal Khan • https://bkkaggle.github.io
gcloud compute instances create gans --zone="us-west1-b" --image-family="pytorch-latest-cu100" --image-project=deeplearning-platform-release --maintenance-policy=TERMINATE --accelerator="type=nvidia-tesla-v100,count=1" --metadata="install-nvidia-driver=True" --preemptible --boot-disk-size="100GB" --custom-cpu=8 --custom-memory=16
gcloud compute ssh gans
-
sudo apt-get update -
sudo apt-get upgrade -
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -
bash Miniconda3-latest-Linux-x86_64.sh -
follow instruction and choose to add to your path
git clone https://github.com/bkkaggle/gans.git
- install https://marketplace.visualstudio.com/items?itemName=rafaelmaiolla.remote-vscode on vscode
- cmd-shift-p and type
Remote: Start Server gcloud compute ssh gans --ssh-flag="-R 52698:localhost:52698"- run
sudo apt -y install ruby && sudo gem install rmateon vm - to edit a file run
rmate path/to/file
conda env create -f environment.ymlconda activate ganspip install future
- get cuda version
nvcc --version - install pytorch
conda install pytorch torchvision cudatoolkit=10.0 -c pytorch
git clone https://github.com/NVIDIA/apexcd apexpip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./
vim ~/.kaggle/kaggle.json- paste in api key
{"username":"[USERNAME]","key":"[API_KEY]"}
wget https://github.com/gdrive-org/gdrive/releases/download/2.1.0/gdrive-linux-x64mv gdrive-linux-x64 gdrivechmod +x gdrive- authenticate with
./gdrive about
There is a Tensorflow version and a Pytorch version
There is no currently available pretrained model for pytorch, so the easiest way to get started is to use my colab notebook
Train on your own dataset as shown in the repository. There currently aren't any pretrained pytorch models.
-
Clone the pytorch repo
git clone https://github.com/znxlwm/UGATIT-pytorch.git -
Download the dataset of your choice
- cat2dog
kaggle datasets download -d waifuai/ugatit-cat2dog-pretrained-model
- cat2dog
-
Unzip the dataset
-
Create a dataset folder under
/datasetfor your dataset (If you're using a pretrained model, the name of the folder must match the name of the saved checkpoint, e.g. cat2dog). -
Create subfolders
testA,testB,trainA, andtrainBunder your dataset's folder. Place any images you want to transform from a to b (cat2dog) in thetestAfolder, images you want to transform from b to a (dog2cat) in thetestBfolder, and do the same for thetrainAandtrainBfolders if you want to train your own GAN. -
Train the model
python main.py --dataset cat2dog --phase train. Add--light=Trueif you get OOM errors.
I'm using the official pytorch implementation.
Train on your own dataset, or use a pretrained model
Follow the instructions in the README:
git clone https://github.com/NVlabs/FUNIT.git
Use the authors animal face dataset and follow the instructions here or substitute it with your own.
Download the pretrained model from Google drive
./gdrive download 1CsmSSWyMngtOLUL5lI-sEHVWc2gdJpF9mkdir pretrainedmv pretrained.tar.gz pretrained/tar -xf pretrained.tar.gz
Run the model as shown in the README:
python test_k_shot.py --config configs/funit_animals.yaml --ckpt pretrained/animal149_gen.pt --input images/input_content.jpg --class_image_folder images/n02138411 --output ./output.jpg
Point --input to the image you want to transform, --class_image_folder to a folder with a few images of the class you want to transform the input image to, and --output to the path for the output image.
I'm using this repository's CycleGAN implementation. It also contains instructions for how to use CycleGAN on multiple datasets.
To install, follow the instructions in the repository, or use my copied instructions here.
git clone https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix
This repository uses a few custom pip packages, install them with
pip install -r requirements.txt
Download one of the official datasets with:
bash ./datasets/download_cyclegan_dataset.sh [apple2orange, orange2apple, summer2winter_yosemite, winter2summer_yosemite, horse2zebra, zebra2horse, monet2photo, style_monet, style_cezanne, style_ukiyoe, style_vangogh, sat2map, map2sat, cityscapes_photo2label, cityscapes_label2photo, facades_photo2label, facades_label2photo, iphone2dslr_flower]
Or use your own dataset by creating the appropriate folders and adding in the images.
- Create a dataset folder under
/datasetfor your dataset. - Create subfolders
testA,testB,trainA, andtrainBunder your dataset's folder. Place any images you want to transform from a to b (cat2dog) in thetestAfolder, images you want to transform from b to a (dog2cat) in thetestBfolder, and do the same for thetrainAandtrainBfolders.
Download one of the official pretrained models with:
bash ./scripts/download_cyclegan_model.sh [apple2orange, orange2apple, summer2winter_yosemite, winter2summer_yosemite, horse2zebra, zebra2horse, monet2photo, style_monet, style_cezanne, style_ukiyoe, style_vangogh, sat2map, map2sat, cityscapes_photo2label, cityscapes_label2photo, facades_photo2label, facades_label2photo, iphone2dslr_flower]
Or add your own pretrained model to ./checkpoints/{NAME}_pretrained/latest_net_G.pt
python train.py --dataroot ./datasets/horse2zebra --name horse2zebra --model cycle_gan
Change the --dataroot and --name to your own dataset's path and model's name. Use --gpu_ids 0,1,.. to train on multiple GPUs and --batch_size to change the batch size. I've found that a batch size of 16 fits onto 4 V100s and can finish training an epoch in ~90s.
Once your model has trained, copy over the last checkpoint to a format that the testing model can automatically detect:
Use cp ./checkpoints/horse2zebra/latest_net_G_A.pth ./checkpoints/horse2zebra/latest_net_G.pth if you want to transform images from class A to class B and cp ./checkpoints/horse2zebra/latest_net_G_B.pth ./checkpoints/horse2zebra/latest_net_G.pth if you want to transform images from class B to class A.
python test.py --dataroot datasets/horse2zebra/testA --name horse2zebra_pretrained --model test --no_dropout
Change the --dataroot and --name to be consistent with your trained model's configuration.
from https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix: The option --model test is used for generating results of CycleGAN only for one side. This option will automatically set --dataset_mode single, which only loads the images from one set. On the contrary, using --model cycle_gan requires loading and generating results in both directions, which is sometimes unnecessary. The results will be saved at ./results/. Use --results_dir {directory_path_to_save_result} to specify the results directory.
For your own experiments, you might want to specify --netG, --norm, --no_dropout to match the generator architecture of the trained model.
To install, follow the instructions in the repository, or use my copied instructions here.
git clone https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix
This repository uses a few custom pip packages, install them with
pip install -r requirements.txt
Download one of the official datasets with:
bash ./datasets/download_pix2pix_dataset.sh [cityscapes, night2day, edges2handbags, edges2shoes, facades, maps]
Or use your own dataset by creating the appropriate folders and adding in the images. Follow the instructions here.
Download one of the official pretrained models with:
bash ./scripts/download_pix2pix_model.sh [edges2shoes, sat2map, map2sat, facades_label2photo, and day2night]
Or add your own pretrained model to ./checkpoints/{NAME}_pretrained/latest_net_G.pt
python train.py --dataroot ./datasets/facades --name facades_pix2pix --model pix2pix --direction BtoA
Change the --dataroot and --name to your own dataset's path and model's name. Use --gpu_ids 0,1,.. to train on multiple GPUs and --batch_size to change the batch size. Add --direction BtoA if you want to train a model to transfrom from class B to A.
python test.py --dataroot ./datasets/facades --direction BtoA --model pix2pix --name facades_label2photo_pretrained
Change the --dataroot, --name, and --direction to be consistent with your trained model's configuration and how you want to transform images.
from https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix: Note that we specified --direction BtoA as Facades dataset's A to B direction is photos to labels.
If you would like to apply a pre-trained model to a collection of input images (rather than image pairs), please use --model test option. See ./scripts/test_single.sh for how to apply a model to Facade label maps (stored in the directory facades/testB).
See a list of currently available models at ./scripts/download_pix2pix_model.sh
This repository is still a work in progress, so if you find a bug, think there is something missing, or have any suggestions for new features, feel free to open an issue or a pull request. Feel free to use the library or code from it in your own projects, and if you feel that some code used in this project hasn't been properly accredited, please open an issue.
- Bilal Khan - Initial work
This project is licensed under the MIT license - see the license file for details
This project contains code from: (https://github.com/taki0112/UGATIT), (https://github.com/NVlabs/FUNIT), (https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix)
This README is based on (https://github.com/bkkaggle/grover) and (https://github.com/rish-16/gpt2client)