Tang Xiaoou focused on artificial intelligence vision technology during his lifetime, and he cultivated a large number of AI talents during his teaching at the Chinese University of Hong Kong.
Tang Xiaoou, founder of SenseTime and artificial intelligence scientist.
Text |Liu Yiqin
Edit |Xie Licheng
On the evening of December 16, artificial intelligence company SenseTime released an obituary, Tang Xiaoou, founder of SenseTime Technology and artificial intelligence scientist, died at 23:46 on December 15, 2023, at the age of 55.
Tang Xiaoou was born in Anshan, Liaoning Province in 1968, graduated from the University of Science and Technology of China in 1990, graduated from the University of Rochester in 1991 with a master's degree, first came into contact with artificial intelligence in 1992, and graduated from the Massachusetts Institute of Technology with a doctorate in 1996. After returning to China after graduating with a Ph.D., Tang Xiaoou taught in the Department of Information Engineering of Chinese University of Hong Kong, founded the Multi-** Laboratory of Chinese University of Hong Kong (the predecessor of SenseTime) in 2001, and worked at Microsoft Research Asia from 2005 to 2007 as the director of the Visual Computing Group.
Since December 2008, Tang Xiaoou has served as the director and researcher of the Multi-level Integrated Technology Research Office of Shenzhen Institute of Advanced Technology. Tang Xiaoou also served as the chairman of a number of important international conferences, including ICCV and CVPR, the top international conferences on computer vision, and won the CVPR 2009 Best ** Award, which was the only Asian ** award in the 25-year history of CVPR at that time. In January 2023, Tang Xiaoou became a member of the National Committee of the Chinese People's Political Consultative Conference.
In 2014, Tang Xiaoou established SenseTime, an artificial intelligence company.
SenseTime was once the world's most valuable AI unicorn company, with 10 rounds of financing completed in seven years of establishment, with a pre-IPO valuation of about $12 billion. On December 30, 2021, SenseTime was listed on the Hong Kong Stock Exchange, with the latest market capitalization of HK$42.2 billion (about US$5.4 billion).
On the 2022 Forbes Hong Kong Rich List, Tang Xiaoou ranks 16th with a net worth of $6 billion.
According to the company's financial report and prospectus, SenseTime has been in a loss-making state for the past few years. From 2018 to 2022, after deducting changes in the fair value of preferred shares, SenseTime's adjusted losses were 1500 million yuan, 11600 million yuan, 8800 million yuan, 14200 million yuan and 4.7 billion yuan. In the first half of 2023, the adjusted net loss was 23900 million yuan.
From 2018 to 2022, SenseTime's revenue was 18500 million yuan, 30300 million yuan, 34500 million yuan, 4.7 billion yuan, 3.8 billion yuan. For the first time in 2022, there was a decline in revenue. In the first half of 2023, SenseTime's revenue will be 143.3 billion yuan, a year-on-year increase of 13%。
In 2022, SenseTime had a downsizing of staff, with 5,098 employees as of December 30, 2022, a decrease of 1,015 from the same period in 2021.
Tang Xiaoou's mission for SenseTime is to "adhere to originality and let artificial intelligence lead human progress". In the exploration of AI technology, SenseTime has seized the opportunity due to its founder's technical acumen, but also faces certain challenges. He made his last public speech in July, speaking about his student experience.
Xiaogang Wang studied in Xiaoou Tang's laboratory during his master's degree and returned to the laboratory in 2009 after graduating with a doctorate. He developed the DeepID face recognition technology, which made the accuracy of face recognition made by machines exceed that of human eyes for the first time. In addition, Wang Xiaogang led the team to defeat Google in 2015 and won the first large-scale visual identity challenge in the history of Chinese scholars, ImageNet, world championship.
He Kaiming studied under Tang Xiaoou during his master's and doctoral degrees. He developed deep residual networks (resnets), a popular architecture in the field of computer vision, which solved the problem of gradient transfer in deep networks and became the basis for large model series such as GPT. Tang Xiaoou said: "He Kaiming made the network deeper, and Google widened the entrance to the network, deep and big, so that today's big model has been achieved." ”
Lin Dahua studied under Tang Xiaoou at the University of Hong Kong Chinese, and he led the initiation of OpenMMLAB, the most influential open source project for visual algorithms in the world.
Tang Xiaoou trained many professionals in the field of AI during his time at Chinese University of Hong Kong, which is also known as the "Whampoa Military Academy" in the field Chinese of AI in China.
The following is Tang Xiaoou's official obituary released by SenseTime: