The multifaceted applications of SenseTime’s AI algorithms – KrASIA

SenseTime is one of China’s leading AI companies. It is chiefly known for its deep learning platform that can recognize and process human faces, images, and text. Plus, it operates a supercomputing center that is equipped with a range of AI technologies. At the moment, SenseTime is expanding into new sectors—smartphones, entertainment, and automobiles. Additionally, the company has developed large-scale infrastructure projects with industry leaders in e-commerce (Alibaba and Suning), telecommunications (China Mobile), and real estate (Vanke and Sunac).

As one of the leading AI developers in China, SenseTime has repeatedly wooed investors. It completed 12 rounds of financing and raised more than RMB 35 billion (USD 5.22 billion) before its IPO in Hong Kong. On December 30, 2021, when the company’s shares started to trade on the Hong Kong Stock Exchange, it raised another HKD 5.775 billion (USD 735.9 million) with a market capitalization of HKD 140 billion (USD 17.84 billion). In 2020, SenseTime was the largest AI company in Asia by revenue.

Smart city technology

Since 2019, smart city development has been one of SenseTime’s main sources of revenue. In early 2021, it was the recipient of many smart city projects, leading to a 240% year-on-year increase in revenue.

SenseTime’s AI platforms can simplify complex urban management processes, automating them in the background. Even though the user may not realize the complexity of the tasks being processed, it takes months for a team of engineers to implement a single algorithm. But different scenarios may require different automation solutions, so SenseTime developed a smart city AI hub equipped with face and vehicle recognition, crowd analysis, and event detection tools. These hubs are now commonly found in cities across China.

Shanghai’s Changning District began to use AI for road management in 2020. Multiple algorithms that addressed various scenarios were created, and 1,000 cameras were transformed into sensors that could detect problems such as illegal parking or flooded roads. The sensors sent alerts to administrators, allowing them to become aware of and solve problems more efficiently.

SenseTime technology can be seen in cities across China and will soon be applied in other sectors.

State-of-the-art technology and highly skilled staff

Tang Xiao’ou, one of SenseTime’s co-founders, went to the US after graduating from the University of Science and Technology of China in 1990. He obtained a master’s degree from the University of Rochester in 1991 and a doctorate in computer vision from the Massachusetts Institute of Technology in 1992. In June 2014, while teaching at the Chinese University of Hong Kong (CUHK), Tang’s team of researchers at the school launched DeepID, a face recognition algorithm with a 99.55% accuracy rate—a benchmark that was more accurate than the human eye’s 97.53%. He co-founded SenseTime with Xu Li, who earned his doctorate at CUHK, later that year.

So far, SenseTime’s staff have already won awards in 70 global competitions and published 600 academic papers. By the end of 2021, 70% of SenseTime’s staff, or roughly 4,300 people, were R&D technicians. The company has built AI “schools” and “libraries” to process data for deep learning, as well as a large AI computing center in Shanghai that can process 23,600 years’ runtime worth of video content in a single day.

To keep its leading edge, SenseTime works with universities around the world to recruit top talent. It sponsors labs and supports AI research through scholarships to encourage, cultivate, and identify potential recruits. SenseTime frequently hires postgraduates from top Chinese universities and attracts experienced employees from other tech enterprises, including Microsoft, Google, Lenovo, and Baidu.

Pandemic prevention in Shanghai

In March, Shanghai experienced a severe COVID-19 outbreak, with more than 51,000 confirmed cases plus another 503,000 asymptomatic cases. To prevent the infection’s spread, hospitals urgently required a safe, effective way of controlling access to their wards and automatically measuring body temperatures. In May, SenseTime developed a “digital makeshift hospital” system, which was implemented at a temporary medical facility in Lingang, located in southeast Shanghai.

SenseTime installed an access control system that can rapidly measure people’s body temperatures even when they keep their masks on. Visitors can enter the hospital by scanning their ID cards and health QR codes, while the system will check their latest nucleic acid test results.

Promoting traditional Chinese culture

Products inspired by traditional Chinese culture have become increasingly popular among Chinese consumers. In April, SenseTime released a digital mural, The Dunhuang Nine-Colored Deer, based on an original painting in the Dunhuang Mogao Caves, a UNESCO World Heritage Site. By using a phone to scan the deer’s image on a decorative block, users could view a digital replica of the original mural.

SenseTime says that videos and 3D models like the deer’s let viewers access cultural relics in more immersive ways. Their main target for these products is Gen Z, who appreciate the application of augmented reality and AI. The digitization of important cultural objects serves the general public by giving people a way to view them in new formats, introducing traditional culture in a more personal way.

SenseTime faces challenges as it goes global

SenseTime has offices in Hong Kong, Kyoto, Tokyo, Singapore, Riyadh, Abu Dhabi, and Dubai. It plans to expand its businesses in countries such as Japan, Malaysia, the United Arab Emirates, Thailand, and Singapore. The AI developer is aiming to cement itself globally. In March 2020, SenseTime signed an agreement with the East Nippon Expressway Company Limited (NEXCO), a highway operator in Japan, to introduce intelligent video analysis technology in NEXCO’s projects.

Talent shortage

According to the 2020 Global AI Talent Report published by Element AI, there were more than 477,950 AI researchers around the world in 2019. Among them, 188,300 were based in the United States—the largest share. This was followed by India in second place. China had 22,191 AI researchers within its borders, fourth on the list.

AI’s rapid development is made possible by a steady pipeline of technical talent. A company like SenseTime can only retain its competitive advantage by constantly recruiting skilled professionals. China was a latecomer to the AI industry, and its technical personnel in this field are less experienced than most of their foreign counterparts, meaning that there haven’t been enough qualified people to fill the AI research positions in China. For now, recruitment remains a huge challenge for SenseTime and its competitors.

Staying in the lead?

The AI sector is becoming increasingly competitive. SenseTime’s main competitors in China—CloudWalk, Megvii, Yitu, and Hikvision—are also developing smart technologies to eat into SenseTime’s business.

In September 2021, Megvii received approval to list its shares on the Shanghai Star Market. Last year, Yitu withdrew its IPO application in Shanghai, and may be considering going public in Hong Kong. Meanwhile, Hikvision invested RMB 3.878 billion (USD 580 million) in the first six months of 2021 to strengthen its R&D capabilities, eclipsing SenseTime’s investment for the full year. To maintain its leading position in China, SenseTime will have to catch up to these other AI developers.

This article was adapted based on a research report by 36Kr Global’s research team. KrASIA is authorized to translate, adapt, and publish its contents.

Spread the love

Leave a Reply

Your email address will not be published.