Started by Aidan Juma for the CSC1028 "Computer Science Challenges" module, which is managed (wonderfully!) by Dr. John Bustard at Queen's University Belfast.
This project aims to build a foundation for the generation of synthetic training data for object detection using AI art techniques. By starting with a simple low-poly 3D object as a base and using Stable Diffusion with ControlNet, photorealistic variations of the object can be created to train computer vision models on.
The ultimate goal is to develop a general vision system that progressively learns to recognise and track a wider range of objects.
Upon recommendation from one of their co-founders via email, I took part in the Backdrop Build v3 pre-accelerator programme. A-Eye became one of 43 finalists. On April 28, Build v4 is due to start - applications close a week earlier, to my knowledge. I'd totally recommend applying! You can do so here.
All repositories and their related code are distributed under the MIT License. See LICENSE for more information.
Aidan Juma - @aidanjuma - aidan@aidanjuma.dev
Special thanks to:
- Dr. John Bustard for the opportunity to participate in this module - I have really enjoyed it, and have learnt a lot over the past 3 months or so.
- Rapha from Backdrop Build for reaching out to me in the first place; Build v3 was a great experience to put my project out on the internet.
