分类： technology

A groundbreaking advancement in digital human technology has emerged from China, where scientists have constructed the industry’s largest high-precision 3D facial database. This development marks a significant leap toward creating exceptionally realistic virtual humans capable of natural emotional expression and identity recognition.

The research collaboration between Shenzhen Institutes of Advanced Technology (Chinese Academy of Sciences) and Fujian University of Technology addressed a critical bottleneck in 3D facial landmark detection—the scarcity of large-scale, precisely annotated datasets. Previous methods predominantly relied on 2D texture assistance or synthetic 3D faces, resulting in limited generalization capabilities.

At the core of this breakthrough is a novel curvature-fused graph attention network (CF-GAT) architecture that can directly predict facial landmarks from raw point clouds. This innovative approach facilitates a fundamental shift from generic ‘one-size-fits-all’ modeling to truly personalized facial reconstruction.

The research team established a customized 3D/4D facial acquisition system to collect standardized data, ultimately compiling approximately 200,000 high-fidelity 3D facial scans. The comprehensive database system encompasses multiple specialized datasets including multi-expression 3D faces, standardized 3D facial landmarks, high-precision 3D human bodies, and dynamic 4D facial expressions.

According to corresponding author Song Zhan, these databases now form the foundational infrastructure for humanoid robot development, enabling high-fidelity perception, expression modeling, and behavior generation. The technology promises to revolutionize human-computer interaction by creating more natural and intelligent interfaces.

The study, recently published in IEEE Transactions on Circuits and Systems for Video Technology, anticipates applications extending beyond digital humans to data-driven large-model humanoid robot systems, potentially transforming how humans interact with artificial entities across various sectors.

KUNSHAN, China — A groundbreaking demonstration in southeastern China offered a compelling vision of urban air mobility’s future. At a low-altitude flight test facility in Kunshan, approximately 60 kilometers west of Shanghai, Chinese aviation company AutoFlight showcased its massive Matrix aircraft—a 5-ton electric vertical takeoff and landing (eVTOL) vehicle recognized as China’s largest electric aircraft to date.

The Matrix prototype, developed since AutoFlight’s 2017 founding, represents a significant engineering achievement with its substantial dimensions: a 20-meter wingspan, 17.1-meter length, and 3.3-meter height. The electric aircraft boasts capacity for up to 10 passengers and can operate for one hour on a single charge.

During a recent demonstration observed by The Associated Press, the colossal eVTOL executed a flawless performance. After meticulous pre-flight checks, the aircraft ascended smoothly from a helipad, generating noticeably less noise than conventional helicopters. The Matrix completed two circuits around the test facility before executing a perfect landing after approximately 10 minutes of flight.

Despite the successful demonstration, significant regulatory challenges remain before such aircraft can transform into operational air taxis. AutoFlight Senior Vice President Steven Yang acknowledged the uncertainty surrounding commercialization timelines, noting the company anticipates securing crucial type certification from regulators by 2027—confirming the aircraft’s compliance with safety standards. Additional operator certifications would subsequently be required before passenger transport can commence.

The Chinese eVTOL landscape continues evolving, with Guangdong-based EHANG having already received certification for commercial passenger services, though operational implementation remains pending. Industry experts emphasize that beyond aircraft certification, developing supporting infrastructure represents a critical hurdle. The emerging ‘low-altitude economy’ already demonstrates practical applications through drone-based food delivery services in cities like Shenzhen.

Gary Ng, senior economist at Natixis Corporate and Investment Banking, highlighted various challenges including safety assurance, infrastructure development, and logistical coordination. ‘The entire ecosystem surrounding this technology remains underdeveloped currently,’ Ng observed, projecting that at least three additional years will be necessary before viable commercial operations emerge.

While the Matrix demonstration makes electric air transportation appear increasingly feasible, Yang emphasized that realizing this future requires collective effort: ‘We truly believe it will happen, but this is not AutoFlight’s job alone—it’s about developing the entire ecosystem.’

Alibaba Group has intensified its artificial intelligence development strategy following the departure of two prominent technical leaders from its Qwen large language model team. The company’s Chief Executive Officer Eddie Wu confirmed in an internal communication that Chief Technology Officer Zhou Jingren will assume enhanced responsibilities in advancing Alibaba’s signature AI model.

The leadership transition follows the resignation of Lin Junyang, a core technical leader who announced his departure via social media with a brief message: “me stepping down. bye my beloved qwen.” Yu Bowen, head of post-training for the Qwen model, subsequently also announced his exit from the company.

Lin, recognized as one of Alibaba’s youngest P10-level technical leaders at age 33, played a pivotal role in developing the Qwen3-Max model featuring over one trillion parameters and later guided the release of the Qwen3.5 series of smaller models.

According to internal sources, the restructuring emerged as Alibaba’s Qwen initiative evolved from a foundational model project into a group-level strategic priority. The transformation necessitated organizational adjustments that ultimately led to Lin’s decision to resign.

In response to these developments, Alibaba is establishing a foundation model support group jointly coordinated by CEO Wu, CTO Zhou, and Fan Yu. This new structure aims to mobilize company-wide resources to accelerate large model development. Wu emphasized that building foundational AI models represents a critical long-term strategy for Alibaba, with continued commitment to open-source approaches and increased investment in artificial intelligence research and development.

The company simultaneously announced plans to recruit additional leading AI researchers to strengthen its base model team talent pool, signaling Alibaba’s determination to maintain competitive advantage in the rapidly evolving artificial intelligence landscape.

The Chengdu High-tech Zone in Sichuan province transformed traditional Lantern Festival celebrations into a spectacular showcase of technological innovation and local talent on Tuesday evening. The event featured a groundbreaking fusion of artificial intelligence and cultural expression that highlighted the region’s growing status as a tech hub.

The centerpiece of the celebration was an immersive light display projected across the 52,000-square-meter LED screens of the Tianfu International Financial Center’s iconic twin towers. This visual spectacle, developed in collaboration with Kuaishou’s advanced AI model Keling, unfolded across seven thematic chapters that blended holiday motifs with digital artistry. The 218-meter towers served as a dynamic canvas for displaying slogans from nearly 50 leading enterprises within the zone, turning corporate messaging into an artistic experience.

Beyond the visual spectacle, the event featured an innovative musical performance where employees from the zone’s electronic information, biomedicine, digital economy, and future technology sectors debuted six original compositions. These works, created through human-AI collaboration, drew inspiration from both professional experiences and daily life, demonstrating how artificial intelligence can enhance creative expression.

The celebration effectively demonstrated how traditional festivals can serve as platforms for technological demonstration and talent recognition, while simultaneously promoting local enterprises and innovative capabilities to a broader audience.

China has declared an accelerated push toward technological self-reliance as a cornerstone of its national development strategy. Science and Technology Minister Yin Hejun outlined a comprehensive roadmap for the 2026-2030 period following the opening session of the National People’s Congress, positioning scientific innovation as critical to China’s transformation into a global technology leader.

The strategic blueprint emphasizes original and pioneering research across the entire innovation chain. Minister Yin highlighted intensified focus on strengthening basic research infrastructure while advancing cutting-edge domains including artificial intelligence, quantum technology, biomanufacturing, and nuclear fusion. The approach marks a significant shift from conventional research conversion models toward integrated planning of technological and industrial innovation.

Technological advancement is positioned as the driving force behind new quality productive forces and modernized industrial systems. The ministry plans to enhance coordination across China’s national innovation ecosystem, particularly leveraging three major international science hubs: Beijing-Tianjin-Hebei, Yangtze River Delta, and Guangdong-Hong Kong-Macao Greater Bay Area. These regions will serve as primary innovation sources, industry leadership centers, and talent attraction clusters.

Substantial infrastructure development accompanies the strategy, including construction of major scientific facilities, establishment of world-class academic journals and databases, and reforms to science evaluation systems. While emphasizing domestic capability building, the plan also commits to international scientific cooperation through leadership in major global research initiatives.

The commitment follows demonstrated progress in China’s innovation capacity. The nation climbed to 10th position in the global innovation index, with 2025 research investment exceeding 3.92 trillion yuan ($570 billion). A landmark achievement saw basic research funding reach 7.08% of total R&D expenditure, surpassing the 7% threshold for the first time. These investments have yielded breakthroughs in open-source large models, semiconductor chips, and pharmaceutical innovations.

China has achieved significant technological independence in blockchain development through groundbreaking domestic innovations, according to statements made by a national legislator during the National People’s Congress sessions in Beijing.

Dong Jin, deputy to the 14th National People’s Congress and director of the Beijing Academy of Blockchain and Edge Computing, revealed that China has successfully developed and deployed indigenous blockchain technologies across multiple critical sectors. This represents a substantial shift from the pre-2019 era when most blockchain applications in China relied exclusively on foreign technologies without domestic core capabilities.

The breakthrough came following a strategic initiative launched by the Beijing municipal government in late 2019, which assembled hundreds of top scientists worldwide to establish a dedicated research institute. Through collaborative efforts, Dong’s team developed two revolutionary technologies: the world’s first software-hardware integrated blockchain operating system, featuring 3 million lines of completely open-source code, and the world’s inaugural 96-core dedicated acceleration chip specifically designed for blockchain applications.

These innovations have produced remarkable performance improvements, with the specialized chip enhancing blockchain processing capabilities by fiftyfold while successfully addressing the computational bottlenecks that previously constrained large-scale blockchain networks. The development marks a significant milestone in China’s digital infrastructure advancement, providing what Dong described as a ‘Chinese core’ for the nation’s technological ecosystem.

Currently, these homegrown blockchain technologies have been implemented across 16 central government departments and 27 state-owned enterprises. In practical application, China’s taxation system now processes tens of billions of invoices annually through this domestically developed blockchain, ensuring complete authenticity, trustworthiness, and traceability for every transaction.

The expansion of these applications has culminated in the establishment of a comprehensive national blockchain network that safeguards China’s high-value core data related to cross-border trade and global payment systems, positioning the country at the forefront of blockchain technology implementation and security.