Computer Vision News - July 2020

Trevor Darrell 46 With all this pressure, and so many offers and proposals coming your way, how do you cope with the workload? I think one of the beauties of being a researcher is that there are few or no boundaries. There’s nobody to tell me what not to work on. In AI, the problems are so broad that one day I could look at optic flow in computer vision, the next day I could look at object recognition, the next day a vision and language problem, and then I might talk to someone about a robotics perception problem. It’s exciting because there is so much progress. Techniques are advancing rapidly and letting us make an impact acrossawiderangeof topics. There’sno safe spot that’s just my little territory. It’s a big competitive field right now. Not having walls is sometimes a blessing but also sometimes adds to the stress. Opportunity creates this workload. When someone with a top background comes to you and says, “I have funding, can I work with you?” it’s not natural to say no. It’s like walking into a bakery - evolution didn’t equip us with the tools to say no to that cupcake! Our response is: “What a great idea!” But you have to very deliberately stop and think, what is my bandwidth? How much time do I have for everything and where will I make the biggest impact? All of us are struggling with that and learning as we go. In our magazine, we try to show the human side behind research, with an aim to improve communication in our community. I’ve noticed that sometimes there can be a great paper, but it goes on a shelf and there is no progress with it because nobody knows about it. What do you think can be done to improve this situation? That’s a really big question. I have a lot of ideas and most of them are too radical for my friends and colleagues. "Results!" Best of CVPR 2020