Everything you need to know about the AI behemoth
SenseTime, originally an academic project, has become a prominent force in the artificial intelligence (AI) industry. It is one of the most valuable AI firms globally, renowned for its facial and image recognition and autonomous driving technologies.
SenseTime is headquartered in Hong Kong. It has built an impressive client and partner roster, including MIT, Honda, and Alibaba. But what makes SenseTime a key player in the AI industry, and in what ways is it shaping the technological landscape of tomorrow? This article aims to answer these questions.
How SenseTime became a big name in AI
Founded in 2014 by Tang Xiao’ou and Xu Li, SenseTime quickly became a major player in AI development, especially in computer vision. The company’s rise in AI was marked by its prominent role in major AI conferences and success in the ImageNet competition in 2015 and 2016.
What is a vision model?
Vision models interpret image data.
The company’s advancements have been somewhat overshadowed by controversies. SenseTime has been accused of assisting the Chinese government in the development of facial recognition and other technologies used in China’s internment camps in the Xinjiang Uyghur Autonomous Region.
This involvement led to an international backlash, including sanctions from the US. These sanctions were based on allegations that SenseTime’s equipment was used in the suppression and human rights abuses of Uyghurs. As a result, the company was banned from doing business in the US and forbidden from using American components, software, or hardware in its products.
In 2022, the United Nations Human Rights Office of the High Commissioner released a report stating China’s systematic abuses and detention against predominantly Muslim groups may amount to crimes against humanity. The Chinese government reportedly uses SenseTime’s ethnicity-detecting AI, along with software from other companies, to surveil and detain Uyghurs in Xinjiang.
More recently, SenseTime has faced two major setbacks. Claims of financial improprieties and the death of SenseTime’s founder caused the company’s stocks to tumble in 2023.
Grizzly Research, a US market research firm that analyzes publicly traded companies, reported SenseTime inflated its revenues by using “round-tripping schemes.” Grizzly Research stated SenseTime or one of its intermediaries would provide funds to its customers. The customers would then use the funds to purchase goods through SenseTime, but those goods may never be received. SenseTime denied the claims in a filing to the Hong Kong Stock Exchange.
Additionally, SenseTime announced its founder Tang Xiao-Ou passed away from an undisclosed illness (Chinese) on December 15, 2023. His passing came just weeks after Grizzly Research released its report.
SenseTime recently broadened its focus in recent years, embracing diversification in its products and partnerships. This shift was marked by the launch of its first consumer product, SenseRobot. SenseTime has also entered into international partnerships, focusing on smart city and digital tourism projects in Saudi Arabia.
Beyond traditional AI: SenseTime’s journey into artificial general intelligence
SenseTime primarily focuses on pioneering artificial general intelligence (AGI) through its SenseNova Foundation Model Sets. AGI goes beyond traditional AI’s task-specific focus, aiming to replicate the human brain’s ability to learn and solve problems. Its hallmark is the capacity to comprehend, learn, and navigate diverse tasks and adapt seamlessly to new scenarios.
The SenseNova platform plays an essential role in this strategy. It provides custom API interfaces and Model-as-a-Service solutions to its industry partners. This integration of SenseNova allows diverse data collection, real-world testing, and refinement of AGI models with feedback loops from businesses using SenseNova. This improves AGI algorithms and expands the scope of SenseTime’s AGI across different sectors.
Source: SenseTime
SenseTime’s expansive AI suite and its capabilities
A closer look at the SenseNova Foundation Model Sets shows they have different capabilities. A standout example is SenseChat, a large-scale language model fine-tuned for the Chinese milieu with billions of parameters. SenseChat’s ability to handle Chinese text makes it a handy tool for jobs, including programming help, health consultations, and reading PDF documents.
SenseTime’s vision extends beyond SenseChat. Under the umbrella of SenseNova, the company also introduced generative AI models and applications. SenseMirage can turn text into images, while SenseAvatar creates digital human avatars. SenseSpace and SenseThings focus on 3D content generation.
These products produce realistic images and avatars and detailed 3D scenes. This highlights SenseTime’s strong potential in the Metaverse and mixed reality areas.
Ernie: Baidu’s ChatGPT competitor explained
Baidu’s Ernie 4.0 comes as China’s formidable answer to the AI race
SenseAuto: How SenseTime is transforming the automotive experience
SenseTime’s expansion into the automotive sector is marked by the introduction of the SenseAuto platform. The SenseAuto platform is a holistic integration of intelligent cabins, intelligent driving, and AI cloud solutions to revolutionize the automotive experience to be safer and more enjoyable.
SenseAuto Cabin
SenseAuto Cabin subsystem features advanced facial verification, vision tracking, head posture recognition, motion recognition, and object recognition. These contribute to a comprehensive AI solution for intelligent cabins, enhancing security, smart interaction, and personalization to improve the end-user experience. The SenseAuto Cabin includes these key technologies:
- The Driver Monitoring System (DMS) keeps the driver safe by watching for signs of being tired, not paying attention, or other risky behaviors. It uses smart AI to check if they’re sleepy, see where they’re looking, notice if they’re distracted, and find other unsafe actions.
- The Occupant Monitoring System (OMS) focuses on the safety and comfort of all cabin occupants. The OMS includes child and CRS status monitoring, emotion analysis, left-behind object detection, gesture identification, pet presence monitoring, and seat occupancy monitoring.
- The In-Vehicle Infotainment (IVI) integrates SenseTime’s advanced SenseMARS augmented reality technologies. The IVI system improves in-cabin visuals, providing AR-based high-precision navigation, beauty camera effects, and smart albums.
Both DMS and OMS contribute to personalizing the in-cabin experience. The subsystems adjust settings like climate control, seating, and infotainment based on detecting the occupants’ state and preferences.
SenseAuto ADAS
SenseAuto ADAS (Advanced Driver Assistance Systems) includes various driving and parking functionalities, including:
- Pilot driving features like automatic emergency braking, lane-keeping help, blind spot checking, and early warning systems make driving safe.
- Parking solutions to park the car automatically. You can even call the car from far away. It remembers where to park in big parking lots. It uses special sensors and cameras to handle different types of parking spaces, even if they don’t have lines.
- Sensing and perception algorithms like monocular 3D sensing, multi-ocular BEV pre-fusion sensing, and laser scanning point-cloud sensing make driving easier. These algorithms cover traffic participants, obstacles, road geometry, and more, ensuring safety and efficiency in diverse driving scenarios.
Source: SenseTime
SenseCore: The powerhouse behind SenseNova
Training AI models requires powerful graphics processing units (GPUs). SenseTime’s formidable AI infrastructure, SenseCore, is equipped to meet this demand with its 27,000 GPUs and a computational power of 5,000 petaflops. Even budget-friendly Chromebooks can access these advanced models and technological strides, all thanks to the cloud-based nature of many AI applications.
The company’s Artificial Intelligence Data Center (AIDC) in Shanghai strengthens the SenseCore AI infrastructure. SenseTime leverages a large data pool from China’s 1.4 billion residents to improve its AI model development. SenseTime has access to this data through a network of partnerships. These collaborations provide AI model training at reduced costs and fuel SenseTime’s R&D.
Diversification into consumer products with SenseRobot
SenseTime, in its bid to diversify its product range, has made strides with its SenseRobot line. The original SenseRobot, an AI-powered robot designed to play Chinese Chess (Xiangqi), marked the company’s entry into consumer products.
Source: SenseTime
At the product launch, Dr. Xu Li, who leads SenseTime as the Executive Chairman and CEO, shared his vision, saying, “Our goal is to create a robot that can physically ‘think’ and ‘act’ with our leading AI technology, bring industrial grade AI technology into every family, and make real interactions with children and older family members.”
Building on this success, SenseTime recently revealed the latest addition to its SenseRobot family, the SenseRobot Go. This one is for kids who are learning to play the game Go. SenseRobot Go provides a library of professional AI Go exercises and offers human-to-machine and online human-to-human gameplay functions.
game pieces. The robot has a digital screen displaying information, suggesting advanced technology for leisure or learning.” src=”https://static1.anpoimages.com/wordpress/wp-content/uploads/2024/01/senserobot.jpg”/> Source: SenseTime
The AI battleground of 2024: SenseTime’s place among tech titans
SenseTime’s story is one of remarkable impact on the AI industry. There may be intense competition in AI during 2024, with companies like SenseTime, OpenAI, Microsoft, and Google pushing AI technology.
In this competitive landscape, Google introduced Bard to go head-to-head with OpenAI’s ChatGPT 3.0. They’re also working on Gemini to compete with ChatGPT 4.0. Meanwhile, Microsoft has the Bing Image Generator in Generative AI. These developments set a high bar for companies like SenseTime to meet and surpass.