David Cai 蔡伟
I am a software engineer at Google, working on image search and
photo search and related computer vision technologies.
Our team has been working on web scale image annotation using deep
neural networks, which is a main component of the search and
organization of the Google Photos.
I received my B.Eng from Tsinghua
University and Master degree from Georgia Institute of Technology. I worked with Professor
Jame M.Rehg at Georgia Tech.
My interests are computer vision, machine learning, and multimedia. My
past research included object tracking, face/object detection and recognition, image retrieval/annotation. I also
worked in Google on large scale image search, visual
browsing and video object segmentation during those years, in close collaboration with Kevin Jing, Henry Rowley and Rahul Sukthankar.
I love technology and enjoy using them to make people's life better. It is for this reason I co-founded VisualGraph, a cloud-based visual shopping platform in 2013. While there I developed computer vision modules for visual search engine. It was acquired by Pinterest in 2014. I believe everyone can do something dazzling as long as you have ambition, big vision, excution and persistence.
In my free time I like reading books and watching NBA games (Rockets
fan for more than a decade). I also do some trading/investing mainly
on US stocks.
I used to be an O'Neil style investor but now I focus on both growth
and value. I really enjoy doing research on public
companies. Their ups and downs could provide tremendous business
insights for me. I believe in the investing precision is much more important than
recall, and the blind diversification is the mother of mediocrity.
2015.5 The new Google Photos launched in Google I/O.
2013.12 VisualGraph was acquired by Pinterest.
2013.11 I joined Google as a Software Engineer.
2013.9 One paper accepted by TMM, one paper accepted by ICCV 2013.
2013.1 I co-founded VisualGraph and worked as an engineer.
2012.11 I will be on leave during academic year 2013.
2012.1 One paper accepted by IJCV.
2011.6 One paper accepted by ICCV 2011.
Page last updated 11/2015