Research problems in CV and ML with GPU computing

I just finished the Udacity course on GPU computing. I work in computer vision, I am excited to read some seminal papers to understand the research opportunities. Please advise.