High-Level Project Summary
Cosmic Mind is an application that aims to produce images with artificial intelligence trained with various images and photographs of space agencies. Cosmic Mind benefits from the wide range of images of space agencies by using the Stable Diffusion module, which enables the artificial intelligence it contains to use reference images. It produces images from scratch, leaving the images it uses to the creativity of users and artificial intelligence. Thus, it aims to attract the attention of the general public to space and technology.
Link to Final Project
Link to Project "Demo"
Detailed Project Description
Cosmic Mind is an application that produces images using artificial intelligence with the parameters it receives. As for how it works, Cosmic Mind includes an AI API trained with images from various space agencies and Stable Diffisioun's own dataset. In addition, thanks to the use of API, users can use artificial intelligence on a server instead of using their own devices, providing both faster and higher quality image production. Thus, users can easily access the application from any device and can produce any image with Cosmic Mind's large data set. As a result, the interest of society, which is the starting point of the Cosmic Mind, is directed to the space and technology. Cosmic Mind makes this with the help of artificial intelligence and open source images provided by the space agencies.
While developing Cosmic Mind, we decided to use the python language, the ease of use of python and the variety of libraries it contains were effective in this decision. Then we decided on the artificial intelligence module we would use and we choose Stable Diffusion with its speed and quality image production. Then we strengthened the artificial intelligence with our own images that we provide from space agencies. Afterwards, we designed a user-friendly interface for ourselves using the PyQt library. Therefore, we have received an easy-to-use, fast and powerful artificial intelligence application. Now let us show you four images we created using our application:
"Jupyter" (Beginner)
"Asteroids floating in Space" (Intermediate)
"Solar System" (Intermediate)
"Betelgeuse" (Advanced)
Space Agency Data
We mainly used NASA's open data source to train artificial intelligence.
The Data we used:
https://www.nasa.gov/pdf/703154main_earth_art-ebook.pdf
https://impact.earthdata.nasa.gov/labeler/
We also used:
https://impact.earthdata.nasa.gov/project/ml.html
Hackathon Journey
First of all, we had a lot of fun and learned as a team during this hackathon. None of us knew until this competition that space agencies released so many datasets to the public. We also learned a lot about artificial intelligence in this process. We all had in mind to do something with artificial intelligence and art so we chose this challenge. We helped each other through this process, so we didn't face many troubles. There were software problems, of course, but we managed to overcome them. Finally, this was our first international challenge, and the Ottawa team helped us a lot. Thanks to each one of them.
References
https://stability.ai/blog/stable-diffusion-public-release
https://www.earthdata.nasa.gov/esds/ai-ml
https://science.nasa.gov/get-involved/art-and-science
https://pypi.org/project/PyQt5/
https://github.com/CompVis/stable-diffusion
https://impact.earthdata.nasa.gov/labeler/
https://www.nasa.gov/pdf/703154main_earth_art-ebook.pdf
Tags
#art, #ai, #software, #space