Selfie2Anime was a successful use of Image-to-Image translation using UGATIT. In this post we learn how to setup a development environment capable of running UGATIT, then train out own variation of a model.
The development of Selfie2Anime has had its ups and downs. This post is an overview of the iterations we’ve made along this journey so far and what the plans are for the future
Learn how to use Conda environments to manage hundreds of different & varying python environments. Clean up your projects and make them more repeatable.
Generating Pokemon cards using a mixture of StyleGAN and RNN to create beautiful & vibrant cards ready for battle!
Run OpenCV 4.0 on Lambda using Serverless framework. Extract images from scanned documents
I’ve been trying to learn more about data science & machine learning recently. Part of the learning process for myself is usually deep diving into interesting projects other people have worked on. I find this very quickly immerses me in …
Automatic coloring and shading of Dune coloring book using manga-style lineart model from deepcolor. Making use of TensorFlow and CGAN
Containerising PyTorch models in a repeatable way. Deploy OpenAI’s GPT-2 model and expose it over a Flask API. Finally deploy it to AWS Fargate container hosting using CloudFormation.
Generate Fake Facebook conversations using OpenAI’s GPT-2 training model and a large number of your existing facebook message data. Perform training using Amazon SageMaker GPU instances and CloudFormation.
Generating Cat images using StyleGAN on AWS SageMaker. Developing a simple and repeatable Data Science pipeline for Generative Adversarial Network outputs.