Computer Vision News 42 CVPR Workshop you can look at howthey’re walking and what their facial expression is, but also look at their electronic health record to see what conditions they’re diagnosed for, when was the last time they fell, what medications they’re on, and so on. That’s good information to include in machine learning models to improve accuracy and performance. Do you have any suggestions for researchers who are looking for a new field of research and would like to make progress in this field? Babak: There are lots of interesting things to work on. What I like about this field is that it is very interdisciplinary, so it’s a lot of collaboration between computer scientists and engineers and clinicians of different fields of practice, but it also brings about lots of really interesting computer science problems. For example, in computer vision, human movement tracking, human gait analysis, pose tracking, facial expression analysis, facial recognition, privacy, fairness, and things like that. A lot of the challenges are in moving technology that is developed in the lab and trying to make it robust and reliable enough to be deployable in real life in the wild applications. That’s often quite challenging but also rewarding to work on to try to get something that is sort of working in the lab but try to make it reliable and robust enough for deployment. Also, be aware of things like fairness. Does the algorithm work equally well for men and women or people with different skin tones? Does it work well for old versus young? If it works well for the faces of young people, does it work well for people who are old and have facial wrinkles and things like that? Is there anything else you think our readers should know? Ehsan: I think healthcare is the new funder. I think the next multibilliondollar company is probably going to be in the healthcare space. Because of that, exploring these interdisciplinary and multidisciplinary fields and their application to healthcare spaces is something to be very much aware of. That’s basically one of the goals of this workshop. In terms of research, I would say one of the things that I think is going to be explored widely in the next couple of years or so is the use of multimodal sensing technologies and foundation models. Building large-scale foundation models with this multimodal sensing that ambient intelligence can provide. That’s the infrastructure, that’s where the data is coming from, and now we can build these large-scale foundation models and use them for downstream tasks and applications of healthcare. In addition to what Babak said, this is another very interesting research field. Beyond that, this is going to be a very interesting workshop with a number of very good speakers in the field. I’m inviting everybody to join us in MICCAI, October 8, in Vancouver, in person, and hopefully, there will be a hybrid component as well.
RkJQdWJsaXNoZXIy NTc3NzU=