Computer Vision News - May 2024

Computer Vision News 32 take another big step. At some point, there is going to be a big step. That is not to say that there isn’t a ton of interesting work left. I want to see so much more work on things like NeRFs and speeding up the 3D reconstruction and recognition process. There are tons of things that I think we can solve with the existing technology, but I also think that there are interesting things that are the next step beyond that. One thing that really bothers me a lot is we kind of have decided that more or less 3-channel RGB images are all there are in the world. I’ve seen stuff coming up, microscopes and satellites, and realizing that, hey, you can do different parts of the spectrum, you can do different bit depths of images, you can do crazy things like event cameras. There are different ways of tackling a similar problem that is how nature seems to solve a lot of problems. I would imagine that that’s important to solving problems in the real world. You look at all these animals that have different perceptual systems, they probably have a different perceptual system for a really good reason to solve their particular problem. The thing that I’m really interested in is tackling that little bit of it. I went to SPIE Photonics West right before the pandemic, and you look at the number of hyperspectral sensing technologies that are commercially available, and they’re starting to really ramp up. I think that’s going to be an area where there’s a lot more interesting things. If you take an interesting problem like, say, plastic sorting or determining if a plant is healthy or not because you’re doing some sort of robotic agriculture, it may very likely be the case that these hyperspectral systems take a problem that, you know, you’re basically stacking up like five GPUs to do the processing with your conventional camera, or you can just do this one weird trick where you use a different camera and the problem turns into something that can be trivially solved with a threshold or something. That, to me, as an engineer, seems like a better solution. [she laughs] I see a wonderful laboratory behind you. Is there anything there you would like to show us in more depth? That’s my partner’s workbench. We split a lab. I don’t have anything super cool in front of me right now. I do a lot of boring manager work – I don’t get all the cool toys! What do you dream of achieving before the end of your career? I have this secret dream that one of Women in Computer Vision