Computer vision and deep learning for on-device AI, trends and state-of-the-arts
Computer vision (CV) is an interdisciplinary scientific engineering field that tackles the problem of how computers gain various levels of understanding from images and/or videos.
In this seminar, I will talk about the recent advances in computer vision, particularly for mobile deep learning (DL) applications. I will begin with the impact of deep learning on the computer vision field, and how it became one of the hottest engineering fields. Then I will go over the latest CV solutions, and observe how the large collection of data and DL approaches are used for deploying the DNN models of the state-of-the-art CV methods on mobile devices. Finally, I will introduce various deep learning-based imaging and computer vision softwares deployed in Samsung’s flagship Galaxy phones, and show how we use the on-device AI scenarios in the phones.
Kihwan Kim is a corporate vice president in the Mobile Communications division (무선사업부) of Samsung Electronics Co., Ltd. He is the head of the Graphics R&D Group and the Visual Solution Lab, working on imaging, computer vision, deep learning (on-device AI), and AR softwares for Samsung’s flagship Galaxy phones. Formerly, he was a principal research scientist in NVIDIA Research, where he worked on the core 3D vision and AI applications for various NVIDIA products such as autonomous driving SDK (DriveWorks), Isaac robotics SDK, VirtualEye, NVSLAM, and gesture recognition framework (DriveIX), etc. He received a Ph.D in Computer Science from the Georgia Institute of Technology in 2011, and B.S. from Yonsei University in 2001.
His research interests span the areas of computer vision, graphics, and machine learning.
More information about his research projects can be found in https://www.kihwan23.com
ID : 897 821 7407
PW : 1nTQDY