Kuaishou’s Award‑Winning AI Research Projects and Their Industry Impact

Kuaishou’s research collaborations have recently received prestigious national technology awards, highlighting the company’s growing strength in industry‑academic partnerships.

The R&D team contributed to two key projects: “Internet video online transcoding, distribution key technology and application” and “Open‑environment adaptive visual perception computing key technology and application.” Both projects won the Chinese Electronics Society Science and Technology Award (first prize) and the 10th Wu Wenjun Artificial Intelligence Technology Award (first prize for scientific progress).

The award criteria emphasize economic and social benefits, requiring not only technical innovation but also rapid, practical deployment that inspires the industry and creates commercial value.

Rooted in real business scenarios, Kuaishou’s research aims to solve practical industry problems, giving it a natural ability to translate research outcomes into production.

One award‑winning project, the online video transcoding and distribution technology, contributed to the industry’s multi‑bitrate live‑streaming standard LAS in 2020. Parts of the code have been open‑sourced, and major domestic CDN providers—including Alibaba Cloud, Baidu Smart Cloud, Huawei Cloud, Kingsoft Cloud, Tencent Cloud, and Wangsu Technology—now support the LAS standard, enabling LAS‑based services.

The second project, adaptive visual perception computing, has been partially applied within Kuaishou’s services. Its model‑compression techniques have been deployed across various deep‑learning models (e.g., de‑blur, super‑resolution, denoising) and have significantly boosted advertising recommendation performance. The same technology earned second place in the 2020 Global Low‑Power Computer Vision Competition, and the related research was published as the paper “UMEC: An Integrated Recommendation Model Compression Framework” at ICLR 2021.

During the fourth Tsinghua‑Kuaishou Research Institute committee meeting, Director Academician Sun Jiaguang emphasized a problem‑oriented, result‑driven approach to ensure research is “usable, manageable, and valuable,” supporting national development goals outlined in China’s 14th Five‑Year Plan and the 2035 vision.

Chief Technology Officer Chen Dingjia echoed this sentiment, noting the importance of leveraging both parties’ strengths to solve technological challenges, cultivate talent, and attract top university graduates to the industry.

Kuaishou continues its frontier research, with 14 papers accepted at CVPR 2021 and many more forthcoming, and it established a postdoctoral work station in December 2020 to nurture talent in cutting‑edge technology fields.