Computer Vision News - February 2020

2 Summary Wo en in Computer Vision 6 Belén Luque Lopez is a computer vision and deep learning engineer at Univrses in Stockholm, Sweden. Belén, tell us about your work. We solve a lot of different computer vision problems, mostly related to 3D and SLAM. We focus on mobile robotics and autonomous driving. Recently, we are developing a product for smart cities, which is actually very interesting. Why is that interesting? It is very interesting for me because I have the opportunity to work on deep learning for that. Previously, I was doing computer vision only. On that project, I got a chance to do some deep learning, which I wanted to do. What is the end result of this project? We are starting now. We are doing some pilots with different cities and different companies. What we want to achieve is that we want to have a cheap device, a cheap camera, like a phone, and gather information about the streets, the state of the streets, the traffic signs, pedestrians, and then have a live feed of that on a map. Everything is on the phone so we are doing edge computing. It means that the computation is in the phone instead of in the cloud. You need a lot of capabilities on the phone itself in order to do all the computing. Yes, but we can also do compression of the networks. That’s what we are "It is not magic. It’s science!"