Computer Vision News - March 2023

39 UM2ii Lab at U of Maryland are in a position where we are piloting at multiple places, I think that would be a success. ” Paul adds, finally: “ I hope that in five years we have funding and we’re publishing papers, but the most important thing is that I wake up every day and I’m having fun, and the people in the lab are having fun too because life is short. The moment this stops being fun, I’ll go and find something else to do. ” learning framework for homogeneous and heterogeneous datasets, ” Vishwa reveals. “ If one group works on liver tumors while another works on pancreas tumors, each can benefit from the other’s work. We don’t need to curate a completely new dataset, but can develop a more generalizable collaborative framework that works on everything. ” The team will also present lifelong learning technology, which is essential because many models are static and do not adapt well to real- life situations with new diseases, scanners, or protocols. This technology enables AI systems to continually learn and integrate new information, including developments like Covid-19. Looking to the future, the team plans to scaleup in the comingmonths and years. Its lab already has a reputation as somewhere synergy is happening, and MDs and PhDs work together in ways that will move the needle. “ How do we move together rather than pulling or pushing each other? ” Paul poses. “ We’re trying to make this multidisciplinary by design. If we can achieve that, many amazing things will happen, and our work will be more impactful. ” And in five years, where will UM2ii be? “ I’d love to see some technology from this lab get deployed into clinical practice , ” Vishwa ponders hopefully. “ It’s a very ambitious five-year goal, but even if we