Computer Vision News - January 2021

Or, you might tell the student, “Let’s think about it. If we were Google, what would we use?” Yes, exactly. I was teaching optical flow theother day, but it really is fundamental, Horn-Schunck method algorithms, this kind of really classical algorithms. Then a student asked me, “What is a state-of- the-art method?” Then I said, “Probably not this method, but I really need to check. First, on which dataset, and the second which algorithm.” There are these kinds of questions. If you have to pinpoint one thing that you learned from your students, what would it be? Sometimes students give me feedback. I really got lots of positive feedback from the students on one lecture. I tried to have a really, really good structure: after I finish some topics, I always go back to the big picture to tell them where these topics come from and where we are going next. For that lecture, I really tried to have a good structure for all the different technologies. That lecture really helped the students. It’s basically a general principle, how to really learn a new topic. Even for myself, how to learn a new topic is really how to understand how to take all the pieces and put them into the whole picture. 24 Women in Computer Vision “Being honest is a very important part of teaching”