ECCV 2020 Daily - Thursday

to develop interesting mobile features for users. Initially the development target was for Photoshop Mix and Fix. Since I’m coming from a background of developing machine learning applications in imaging and computer vision, and the team is good at mobile app development. So naturally, I am thinking whether there is anything that I can contribute to the team by matching the two different expertise. I’m thinking, maybe I should start working out efficient neural networks and work out on-device runtime to do model inference, so that different models can run on mobile devices without the need to transport users' private data to the cloud. Running neural networks on mobile devices is beyond most people’s imagination in 2015. We started there, and then a couple of months later we had an initial prototype! After that, there was a chance to do a new camera application on mobile devices. With a camera app, the most important thing is to make it real-time, and to make it AI-assisted, smart, and intelligent. My early work of running neural network on device actually became one of the fundamental parts of this mobile camera app. It’s one thing to know how to do it, but it seems that you really like what you are doing. Can you share what it is like working for Adobe? Adobe is a very interesting company. It has a lot of people that are passionate about imaging. One of the special aspects about this camera app is actually to try to help people to get creative. We provide a lot of creative effects that, in the past, only professionals could do with Photoshop. But now everybody can share this creativity in the capture space easily. They can take pictures with professional-level effects with just a couple of clicks. Which kind of devices support this app? It is available to both IOS users and Android users. 3 Xin Lu 25 DAILY T h u r s d a y