SOTA (state-of-the-art) model

Hi everyone, I am Devkumar Biswas, a PhD scholar (sAIDE) at IIT Ropar. I am thinking of building my own SOTA (state-of-the-art) model for segmentation or object detection to improve healthcare-related problems. As you know, to achieve a SOTA result, we need to surpass the previously established benchmarks. I have many ideas, but I am not sure where to start. I believe many of us feel the same way- we want to do something impactful but don’t know how to begin. Let’s improve ourselves by sharing thoughts and ideas on SOTA models in the AI era.