SenseTime Leads the Development of China's First National Standard for Digital Humans

SenseTime led the development of China’s first national standard in the field of virtual digital humans— Information Technology - General Technical Requirements for Customer Service-oriented Virtual Digital Humans (GB/T 46483-2025)—which was recently officially released. This standard establishes unified technical specifications and reference guidelines for the design, development, testing, application and maintenance of customer service virtual digital human systems. SenseTime will actively promote the global adoption of Chinese standards, contribute to building a global technical standard system for digital humans, and accelerate the application of digital humans in industries such as finance, communications, education, and e-commerce.

Leveraging over six years of technical expertise, commercial implementation experience, solid industrial research, mature AI operational capabilities and multimodal technical advantages in the digital human field, SenseTime played a pivotal role in formulating the standard, as the core lead and Vice Chair of the working group. Collaborating with more than 30 institutes in the industry, academia, research and application sectors, the team spent two years constructing the standard framework and refining its details.

Participating in the formulation of industry and national standards to break bottlenecks for industry development

This national standard for digital humans defines for the first time the product form, functional boundaries, and performance indicators of customer service digital humans. It serves as a unified specification for enterprises in product design, development and application. It marks the entry of China’s customer service virtual digital human industry into a new stage of standardized, high-quality development.

This is not the first time SenseTime has taken the lead in defining AI digital human technology and implementation standards. In 2023, SenseTime’s "SenseAvatar" AI digital human generation platform passed the first batch of "Trusted Virtual Human Generated Content Management System" evaluations by the China Academy of Information and Communications Technology (CAICT), earning the "L3 Excellent Level" certificate. With capabilities such as technical anti-counterfeiting and content traceability, it set a security baseline for the industry, demonstrating that SenseAvatar meets industry benchmark levels in functional completeness, system trustworthiness, and industrialization capacity.

Continuously unlocking the commercial and social value of digital humans

The newly released national standard systematically specifies the functional and performance requirements for 2D/3D digital human image generation, image driving, visual interaction, voice interaction, emotional interaction, and operation and maintenance, establishing clear quantitative standards for digital human applications in customer service scenarios.

SenseTime’s SenseAvatar fully complies with the standard, covering multiple commercial scenarios. It has helped numerous clients achieve rapid implementation and improved market penetration, becoming a benchmark for standardized industry applications.

Powered by the strong multimodal capabilities of SenseTime’s SenseNova Large Model, digital human technology is evolving rapidly to meet market demand. SenseAvatar is widely applied in multiple commercial scenarios in various customer service roles:

Service-oriented digital humans (Q&A digital humans) are used in finance, cultural tourism, education, and communications. Clients include Bank of China, Agricultural Bank of China, Industrial and Commercial Bank of China, China Construction Bank, Bank of Ningbo, Bohai Bank, Bank of Shanghai, Sanmenxia Cultural Tourism Group, China Mobile, China Telecom, New Oriental, Tsinghua University, TAL Education Group, and Shanghai University.

Live streaming e-commerce digital humans support customer-service scenarios in thousands of live broadcast rooms on Taobao, Meituan, and JD.com.

Image-to-digital-human technology represents another technological breakthrough: users can generate high-quality videos in minutes with just one image and an audio clip. It supports real-person photos, 2D anime characters, and 3D virtual avatars as driving sources, stably producing multilingual dynamic videos of over three minutes with rap-level lip-sync accuracy, while maintaining high consistency in character identity and background style. It is suitable for commercial marketing, media communication, education, and cultural fields.

In addition, SenseTime continues to expand the application boundaries of digital human technology. It created China’s first AI large model-powered digital human based on the ancient sage "Laozi," exploring new paths for modern communication and immersive experiences related to cultural heritage. It also launched the first "AI Digital Human Qian Xuesen" to support aerospace science popularization education, enabling AI technology to unlock greater value in cultural communication and public welfare initiatives.