ChinAI #201: Chinese University of Hong Kong — Cornerstone of China's Computer Vision Scene
The 3rd article in Leiphone's series on A Decade of Deep Learning's Rise
Greetings from a world where…
more people should give the Netflix show “Mo” a chance!
…As always, the searchable archive of all past issues is here. Please please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay support access for all AND compensation for awesome ChinAI contributors).
Feature Translation: Chinese University of Hong Kong as the Whampoa Academy for China’s Computer Vision Ecosystem
Context: Ten years after AlexNet birthed a new era of deep learning, Leiphone’s AItechtalk portal (AI科技评论) is publishing a great series that looks back at this past decade. The third article (link to original Chinese) in this series is about the critical role of Chinese University of Hong Kong (CUHK) in forming China’s computer vision ecosystem. These are well-researched longform pieces (full English translation will end up being around 7,000 words). Below are some of the most absorbing passages from the first half; we’ll save the second half for next week’s issue.
Key Takeaways: In 1997, the Information Processing Center (IPC) of the University of Science and Technology of China (USTC) received an unexpected guest: Tang Xiaoou.
You may know Tang as a cofounder of the AI giant SenseTime, but at the time of this story, he had just finished his PhD at MIT and was about to start as a professor at CUHK, in the Department of Information Engineering. Before going to Hong Kong, he wanted to go back to his undergraduate alma mater (USTC) and have a look around. He makes some connections with people at the IPC and goes back from time to time for academic exchanges.
One of those connections is Li Xuelong, a junior researcher at USTC. Li was intrigued by Tang’s visit, as Tang was dressed in Western-style clothing and spoke with humor.
Tang, just starting out his CUHK career, struggles to recruit people to his “Multimedia Lab” (MMLab) to research facial recognition.
Now, keep in mind that this is the late 1990s/early 2000s. CUHK’s information engineering department is focused on communications, as image recognition and AI are not hot topics. Plus, Tang lacks contacts, funds, and (most importantly) students in Hong Kong. Cantonese is the main communication language and most of the best students planned to go to the U.S. for graduate study.
Desperate for another strategy, Tang invites Li Xuelong to visit MMLab. It opens up Li’s perspective about the field of computer image processing. At USTC, researchers only knew how to submit papers to domestic journals; at CUHK, Li learns about top international conferences like CVPR.
Li becomes the go-between between Tang’s lab and USTC. This allows MMLab to attract talented students like Wang Xiaogang and Tao Dacheng.
Wang was part of a selective class of USTC students (the 00 Class). Basically, only the top three scorers on the gaokao exam in each province applied to be part of this class. He also won the highest prize among USTC students during undergrad. In AI circles, another USTC alum who received this award was Li Deng (who has since gone on to positions like Microsoft’s chief scientist of AI).
Tao and Wang both go to CUHK on two-year master’s degrees, where under Tang’s guidance they build an impressive research environment at MMLab. They each get two papers accepted into CVPR in 2004. The lab as a whole gets 7. For reference, the total number of accepted CVPR papers is only about 300.
Since their MMLab days: Wang got his PhD at MIT’s vision group and then co-founded SenseTime. Tao joins Hong Kong Polytechnic University as an assistant professor and is now the first dean of the JD.com Explore Academy.
Back next week with the full translation. Here are the first two parts of the story: The person who built the Chinese University of Hong Kong into China’s AI Vision Whampoa Academy
ChinAI Links (Four to Forward)
Should-read: America and China Don’t Need to Knock Each Other Out to Win
Jessica Chen Weiss’s much-needed intervention via The New York Times opinion section. Her Twitter thread has some great additional discussion on the op-ed, which argues: “U.S.-China competition risks becoming an end unto itself, fueling overextension abroad and undermining democracy at home. It's not too late for Xi and Biden to find a way back from the brink.”
Should-read: Xi sends supportive message to domestic IT, AI, and new energy sectors in 20th Congress
Ben Jiang, for South China Morning Post, provides a good distillation of the points relevant to AI and other strategic emerging industries in Xi’s message to the 20th Communist Party Congress.
Should-read: Microsoft Research Asia: The “Whampoa Academy” of China’s Internet Industry
This week’s feature translation references CUHK as the “Whampoa Military Academy” of China’s computer vision scene. This academy was a key training ground for elite military leaders who fought in many of China's conflicts in the 20th century. It reminded me of ChinAI issue #37, which used the same reference to describe Microsoft Research Asia’s role in training talent in China’s Internet industry.
Should-read: The Political Reality inside Metaverse Cities
Zhanna Smith, at CSIS, has a useful primer on metaverse platforms, including China and South Korea's efforts to develop state-backed metaverse cities.
Thank you for reading and engaging.
These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.
Check out the archive of all past issues here & please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).
Any suggestions or feedback? Let me know at chinainewsletter@gmail.com or on Twitter at @jjding99