In 2014, the famous Generative Adversarial Network (GAN) technology was born, and since then there’s been a lot of fun face-changing research in the AI world. GitHub has a host of face generators for celebrities, celebrities, supermodels, whatever you want.

Here are five interesting GitHub projects that use GAN, including one-click face swaps, comic book avatas, comic book styles, and more. I’ve covered how GAN works in a history post on GitHub’s Mosaic-removal projects, so check it out if you’re interested.

01. Face generator

This open source project is a new version of the face generator based on StyleGAN2, which can generate the faces of Internet celebrities, celebrities, supermodels, cute children, Hong Kong style beauties, etc., and also can edit the face attributes, such as adjusting the age, adding glasses.

Address: https://github.com/a312863063/generators-with-stylegan2Copy the code

The following are synthetic faces trained using a dataset of Hong Kong beauties:

It can also edit the familiarity of faces, such as regulating smiles:

Change of age:

Adjust gender:

You can even adjust your appearance level:

02. Video face-switching

Face-swapping project Avatarify is taking GitHub by storm. With this technology, you can replace your own face with someone else’s face in real time and perform smoothly in a video conference. The project has garnered 13.8K stars on GitHub to date.

You don’t even need to train the model from scratch, someone has released a client version that supports Windows.

Address: address: https://github.com/alievk/avatarify-python https://github.com/alievk/avatarify-desktopCopy the code

Instead of swapping out celebrity faces, you can swap out dogs:

03. One-click animation style generation

Take a picture and convert it into the hand-painted style works of Miyazaki Hayao, Makoto Shingai and other Japanese masters. This open source project, which is specialized in generating animation images, is very easy to use in field measurement. GitHub does have an open source project where you type in a photo taken on your phone and instantly convert it to a Japanese anime style.

Address: https://github.com/TachibanaYoshino/AnimeGANCopy the code

04. Animation avatar

This open source project is fun, upload or take a picture of yourself, 2 seconds later you can generate your own animation avatar.

Address: https://github.com/minivision-ai/photo2cartoonCopy the code

05. Automatic P chart

Have you ever thought of a scene where you look at an original picture and say, “Help me P a handsome liu-hai, pa!” An image with a handsome haircut is automatically generated.

Yes, there are open source algorithms that can do this. Enter the original image, just add a description of what you want it to look like, and it will do it for you automatically.

Here are some examples:

The first column is the original image input, the second column is mohican Style, the third column is curly hair, the fourth column is African Style, and the last column is pink.

The last row of generated Musk laughs me to death, especially broom head.

Finally, in case you can’t find this article, you can save it by clicking “like”. Welcome to follow the public account GitHub (ggGithub).