Using AI to fuel productivity

SenseTime is a global leader in artificial intelligence and the first unicorn (with market valuation exceeding HK$15 billion) at Hong Kong Science Park under the banner of Hong Kong Science and Technology Parks Corporation. Focused on computer vision and deep learning, SenseTime has become China’s largest AI algorithm provider.

Focus R&D on specific propositions

Currently, the main direction of AI R&D is to make computers more like humans at solving specific cognitive problems, at a faster and more accurate rate. This will help to enhance efficiency and lower costs for various industries, as well as make life easier and more convenient for people.

In 2014, Professor Tang Xiaoau, an outstanding fellow of the Faculty of Engineering of The Chinese University of Hong Kong and founder of the university’s Multimedia Laboratory, joined forces with a group of academics and engineers to co-found SenseTime. Within two to three years, they quickly leveraged its deep learning platform and products to propel the upgrade of industries, in essence constructing China’s deep learning engine and AI ecosystem. They focused primarily on computer vision and deep learning. Their technology has achieved outstanding R&D capabilities and globally-recognized research results, and proved to work well in application.

Broad application of smart image recognition

Using facial recognition to unlock mobile phones or access online transactions are some familiar applications of smart image recognition. SenseTime has extended its use to a much wider field, including facial recognition access control, photo touch-up mobile apps, online financial services identity authentication, autonomous driving solution and surveillance of public areas. The company has acquired more than 400 clients in the fields of public security, finance, mobile communications, mobile internet and robotics. Qualcomm, NVidia, China Mobile, China UnionPay, Xiaomi, OPPO, vivo and Weibo are some of the big names in its blue-chip client list.

Deep learning creates super “machine brain”

SenseTime’s computer vision and deep learning is supported by its proprietary deep learning platform and a deep learning supercomputing center which lets computers learn as a baby would. Based on existing big data, it derives specific rules on different levels as parameters for making judgments and decisions on similar circumstances in the future. As a machine’s capacity and computing power is far superior to a human brain, and a computer doesn’t get tired, it can think more thoroughly and deeply than humans after referencing enough data, and get a job done much faster and better too.

In image recognition, SenseTime uses its deep learning engine to create a “machine brain”. After training it with big data, it can recognise all sorts of graphics, or specific objects or unusual occurrences in an image or footage, at a speed and precision level humans can’t possibly match. It will be able to work out instantly if a network login is authentic, if there are people posing a threat in a public area, or which wrinkles or blemishes on a face to blur out to achieve the most natural and appealing photographs, for example.

Sustained lead on two fronts

SenseTime co-founder and CEO Dr Xu Li said that the company is leading in technology and commerce. On the technology level, he said: “The AI race is a competition for algorithms for deep learning. It reflects the learning speed and quality of the ‘machine brain’. SenseTime has built an unrivalled talent pool over the years in this area, way ahead of our competitors.” As well, SenseTime’s client list reflects that the company has already successfully integrated AI with the commercial understanding of various industry verticals and has helped to propel the productivity of these industries, which in turn fuels the company’s rapid growth.

In recent years, SenseTime has formed partnerships with leading high-tech manufacturers such as Qualcomm, NVIDIA and Honda. The company expects to create more AI products and services to unleash the productivity of industries and societies.

Contact Us

For general enquiry, or questions about HKSTP’s services and support, please enter your email address and query here.