Computer Vision News - November 2021

32 Best of ICCV 2021 reduce that. The other direction would be to give it generative capabilities so that’s basically the direction I’m working on now. ” We have to ask Aishwarya what it is like to be advised b y Yann LeCun . Did she feel a lot of pressure when she first found out? “ Of course! ” she laughs. “ When I went for my NYU interview, I had an open mind about all the professors that I could end up working with, but then I had an interview with Yann, and it went very well. He told me he was planning to take a PhD student and asked, ‘Do you want to be my PhD student?’ I was like, ‘Whaaat?’ Yeah, definitely I was a bit stressed in the beginning! But once I joined and we started working together it went away. ” What is he like to work with? “ He’s very chilled and very easy to talk to. He’s patient and doesn’t push me to work on things that he’s interested in, but more supports me to explore the things that I want to explore. In general, text is not something that he’s used to working with, he’s more of a computer vision person, but we discussed that my interests in the past were mostly text-based, so I would work at the intersection of language and vision. He was happy about that and recognized that there’s a lot to be explored in this area. ” On that subject, Aishwarya tells us she will be attending the 4th Workshop on Closing the Loop Between Vision and Language on Sunday. Recently, Aishwarya has been working at Google AI as an intern on different projects, but still in the same space of image-to-text understanding . “ For my PhD, multi-modal understanding is the kind of space that I’m interested in, ” she says. “ I’m trying to get experienced with many kinds of applications for this work and trying to figure out how I can use the technology to help people at some point. That’s my end goal. ” “He was happy about that and recognized that there’s a lot to be explored in this area…” Aishwarya Kamath