156_NeonFusion_TheFutureOfArt

Inspiration

The main inspiration behind this piece is the lack of artwork that intersects traditional pieces of art and futuristic pieces of art. Futuristic pieces tend to be digitally inspired and traditional pieces are based heavily on culture and history of civilization. We sought to find a middle ground by integrating the past and present to make what we consider would be the future of art.

How I built it

Our work can be divided into two streams AI Stream and Drone Stream: 1) AI Stream -We created and curated a dataset of English letters and Arabic Symbols -We ran the letters through a FCGAN -We ran the letters and symbols through another FCGAN -We used NST to add styling to the previous output 2) Drone Stream -We started with creating hand 3d light painting to test the camera and set an initial benchmark -We attempted to create drone scripts that translate output from GAN to a drone trajectory -We created a drone prototype that is capable of creating 3D Light Paint

-Finally, We combined artwork from the NST and the drone light paint

Challenges we ran into

AI Challenges: -Slow internet resulted in slower use of CoLab for training of neural networks slowing the whole process -Usual syntax and semantic errors needed a lot of debugging Drone Challenges: -The drone integrations were not working properly -Drone and camera had multiple battery issues which led to huge setbacks on our final idea

Accomplishments that I'm proud of

-We managed to create one of the very few pieces of automated drone art (3D Light Paint) that exist today! -We managed to run and train through three neural networks to reach to the art piece -We managed to create our own datasets by scraping google images and manipulating images -We did all of this despite not having an idea or a team before the artathon started. We closed our idea only on Friday morning and found great chemistry and synergies between different team members

What We learned (Team TEK)

In a span of only 30 hours of non stop work we learned more than what we learned in the past year -We managed to not only understand but train two different kinds of neural networks (GAN and NST) -We created a drone prototype that is capable of light painting -We learned the basics of light painting and built up our project on it

What's next for Team TEK (Neon Fusion : The Future of Art)

-Clearing the first and second rounds of Artathon to go to the bootcamp -Improvise and mature our concept better dedicating more time and leveraging received mentorship