Xander Steenbrugge, a machine learning researcher, created an algorithm that turns images of human faces into animals and back again using artificial intelligence generative systems. Currently, the field of imagery created by AI is relatively new for us, and it might sometimes be thrilling. Most of the results, these research labs are developing are additional upgrades to existing technologies.
It seems that every week some research lab rolls out a new, slightly enhanced algorithm or advances a whitepaper to arXiv. Sometimes, we get to see something exciting for Animorphs books’ fans and robotics engineers. Xander Steenbrugge, has developed a similar thing using a series of AI models that turn humans into Animorphs.
As most of the photo style transfer creations are developed using GANs (generative adversarial networks), Steenbrugge also used the same technique. Generative Adversarial Networks use a particular training dataset for learning and then tries to convert a target image into the source the users feed it. For instance, deepfakes are also created using generative adversarial networks.
Steenbrugge calls his image style transfer creations the ‘humanimals,’ but we call all evidently see that they are Animorphs. Steenbrugge stated that he used a dataset containing 15,000 HD animal faces during the process. The data set he used was published with StarGAN v2, which is a newly introduced generative model. He combined that dataset of 15,000 HD animal faces with a set of human faces and used the combined data as the ‘training data.’ He then ran that training data via a different generative model called ‘StyleGAN v2.
The resultant developed by Steenbrugge is this inter-species artificial intelligence dream. This photo style transfer creation changes a woman into an eerie cat, or you may see a man becoming a man-leopard with aviator sunglasses, or a man is turning into a lion, etc. It seems like we are creating thrilling and exciting results every day by using machine learning models, and to me, these results developed by Steenbrugge are very Zootopia.
Steenbrugge, while discussing his experiments with media outlets, told that most scholarly papers publish a particular method with a planned purpose ‘A’ in mind. But as machine learning engineering techniques are so flexible, sometimes it turns that a researcher can also perform B, C, and D using the same procedure. Researchers only need to run it on a different dataset to do the B, C, or D with that model, Xander Steenbrugge added. He says that often this is where all the fun begins.
The above-discussed experiment is part of Steenbrugge’s broader project, namely ‘Neural Synesthesia.’ Steenbrugge develops immersive audiovisual experiences in his Neural Synesthesia project. He uses machine learning systems and AI algorithms to create those experiences and believes that we are entering a new era in digital media. He said that creativity is turning into an interactive practice between machines and humanity. Steenbrugge is of the view that he does not fully develop his works; instead, he co-create them with the help of artificial intelligence systems that he brings to life.
Read next: This AI Meme Generator Is Making More Hilarious Memes Than Humans
It seems that every week some research lab rolls out a new, slightly enhanced algorithm or advances a whitepaper to arXiv. Sometimes, we get to see something exciting for Animorphs books’ fans and robotics engineers. Xander Steenbrugge, has developed a similar thing using a series of AI models that turn humans into Animorphs.
As most of the photo style transfer creations are developed using GANs (generative adversarial networks), Steenbrugge also used the same technique. Generative Adversarial Networks use a particular training dataset for learning and then tries to convert a target image into the source the users feed it. For instance, deepfakes are also created using generative adversarial networks.
Steenbrugge calls his image style transfer creations the ‘humanimals,’ but we call all evidently see that they are Animorphs. Steenbrugge stated that he used a dataset containing 15,000 HD animal faces during the process. The data set he used was published with StarGAN v2, which is a newly introduced generative model. He combined that dataset of 15,000 HD animal faces with a set of human faces and used the combined data as the ‘training data.’ He then ran that training data via a different generative model called ‘StyleGAN v2.
I published my 'humanimals' model & inlcuded a simple Colab notebook to discover your own latent species, have fun Twitterverse😋!— Xander Steenbrugge (@xsteenbrugge) April 29, 2020
Colab notebook: https://t.co/AfTAUIQtsL
Raw model link: https://t.co/GG8N9m43MG pic.twitter.com/UXjumuFzt4
The resultant developed by Steenbrugge is this inter-species artificial intelligence dream. This photo style transfer creation changes a woman into an eerie cat, or you may see a man becoming a man-leopard with aviator sunglasses, or a man is turning into a lion, etc. It seems like we are creating thrilling and exciting results every day by using machine learning models, and to me, these results developed by Steenbrugge are very Zootopia.
Steenbrugge, while discussing his experiments with media outlets, told that most scholarly papers publish a particular method with a planned purpose ‘A’ in mind. But as machine learning engineering techniques are so flexible, sometimes it turns that a researcher can also perform B, C, and D using the same procedure. Researchers only need to run it on a different dataset to do the B, C, or D with that model, Xander Steenbrugge added. He says that often this is where all the fun begins.
The above-discussed experiment is part of Steenbrugge’s broader project, namely ‘Neural Synesthesia.’ Steenbrugge develops immersive audiovisual experiences in his Neural Synesthesia project. He uses machine learning systems and AI algorithms to create those experiences and believes that we are entering a new era in digital media. He said that creativity is turning into an interactive practice between machines and humanity. Steenbrugge is of the view that he does not fully develop his works; instead, he co-create them with the help of artificial intelligence systems that he brings to life.
Read next: This AI Meme Generator Is Making More Hilarious Memes Than Humans
