CFGs & Reinforcement Learning for ML Computer Vision

As my 3rd-year Computer Science dissertation at the University of Oxford, I carried out an investigation of the use of Context-Free Grammars & Markiv Decision Processes to address some pressing issues in the field of AI raster image manipulation. My methods produced positive results, and aligned with my initial goals & motivations, making the project a success.

The code was written entirely in Python, using JSON for data storage purposes throughout. The ML components were built from scratch (namely, Markov Decision Processes, and the typical Reinforcement Learning 'Series' structure) to run in tandem with my image encoding & filter generation technologies, to produce engaging, valuable results.

Unfortunately, I cannot include image results or the full project report yet, since it is still undergoing the marking process for my final degree certification, but I am happy to discuss the project at length if this is something you are interested in!