Keynote Presentations from the 2022 Global AI Technology Conference – First Industrial Vision Frontier Forum

The 2022 Global Artificial Intelligence Technology Conference’s “First Industrial Vision Frontier Forum”, organized by the China Association for Artificial Intelligence and the Hangzhou Municipal Government and hosted by the Hangzhou Zhuoxi Brain and Intelligence Research Institute, was successfully held on November 27 in Hangzhou, Zhejiang. The forum was co‑hosted by Professor Ding Guiguang, Vice Dean and Special Researcher at Tsinghua University School of Software, and Mr. Yan Weipeng, Vice President of JD.com, Chairman of the JD Retail Technology Committee, and Head of JD Retail Technology R&D and Data Center.

The forum invited seven distinguished speakers: Associate Researcher Tao Xian from the Institute of Automation, Chinese Academy of Sciences; Zhao Tiancheng, Director of the Om Research Center at Zhejiang University’s Binjiang Institute; Professor Bai Cong, Associate Professor at Zhejiang University of Technology; Vice President Bao Yongjun of JD.com; CEO Zhou Jile of Qingsoft Microvision; CTO Guo Hongwei of Yuan Tong Technology Co., Ltd.; and Professor Ma Hanjie, Associate Professor at Zhejiang University of Science and Technology’s School of Information and Founder and Chairman of Hangzhou Maquan Information Technology Co., Ltd. They delivered insightful talks covering topics such as industrial AI visual defect detection algorithms, multimodal pre‑training large models, smart meteorology, JD’s digital intelligence practice, third‑generation compound semiconductor intelligent detection, and meta‑imaging.

Yan Weipeng delivered the opening remarks. He stated that industry is a pillar of the Chinese economy, and under policies that strongly support the development of the real economy, data will become a new production factor in manufacturing, driving intelligent transformation and efficiency gains. Industrial vision, as the “eyes” of industry, will shine in data acquisition, recognition, and inspection. He expressed enthusiasm for exchanging ideas on the frontier of industrial vision and sharing the latest technological developments and applications from various perspectives.

Associate Researcher Tao Xian presented on “Latest Advances in Industrial AI Visual Defect Detection Algorithms”. He first outlined the current state, models, and methods of industrial quality inspection, noting that complex real‑world industrial scenes pose great challenges for supervised algorithms. He then introduced progress in unsupervised learning‑based visual inspection and suggested future directions, concluding with a brief overview of his team’s ongoing work and results.

Zhao Tiancheng presented “Multimodal Pre‑training Large Model OmModel and Industrial Vision Application Development”. He introduced the challenges and background of pre‑training large models, described OmModel—a multimodal pre‑training large model that can cover many long‑tail scenarios while significantly reducing deployment costs and accelerating visual algorithm production. He concluded with case studies of vision detection and recognition applications powered by large models, demonstrating industry adoption.

Professor Bai Cong presented “An Initial Exploration of Smart Meteorology Based on Computer Vision”. He discussed the research background, construction of meteorological databases, classification of imbalanced disaster weather satellite images, typhoon prediction and warning, and short‑term rainfall forecasting based on radar data, highlighting frontier research in computer‑vision‑driven smart meteorology. He concluded with several reflections on the field.

Vice President Bao Yongjun presented “Key Technologies Behind JD’s Digital Intelligence Practice and Their Application in the Retail Industry”. He described how new consumer trends drive digital intelligence transformation in retail, and how JD leverages digital intelligence to adapt and lead, achieving intelligent marketing, supply‑chain, and omnichannel transactions, continuously optimizing cost, efficiency, and experience. JD promotes AI innovation across the entire product content lifecycle, using multimodal pre‑training as a technological foundation to advance digital intelligence. He concluded with examples of JD’s digital applications that accelerate retail business growth.

CEO Zhou Jile presented “Opportunities and Challenges of Intelligent Detection for Third‑Generation Compound Semiconductors”. He introduced the background, characteristics, typical application scenarios, and industry deployment of third‑generation compound semiconductors, noting that the new energy vehicle sector is becoming the main market for SiC. He concluded with a comprehensive discussion of the challenges and future trends of SiC technology.

CTO Guo Hongwei presented “Meta‑Imaging Applications and Outlook”. He first introduced the meta‑imaging design concept, which differs from holistic approaches by addressing imaging system bottlenecks. He then explained the principles and innovations of meta‑imaging technology, achieving an intelligent light‑field solution with a large field of view, high resolution, compact size, and aberration‑free performance. He concluded with prospects for the future development and market potential of meta‑imaging.

Professor Ma Hanjie presented “Industrial Artificial Intelligence Cases and Exploration”. He examined the key role of industrial AI in the industrial internet through discussions of technological change, policy demands, and industry analysis. He introduced typical industrial AI techniques such as few‑shot learning algorithms and case studies like an industrial safety governance platform. He concluded by highlighting technical challenges and future research directions for industrial AI.

At the close of the forum, Professor Ding Guiguang thanked the speakers and online audience, expressing hope that the insightful presentations would provide valuable insights and inspiration, and encouraging continued attention to the activities of the China Association for Artificial Intelligence.

The forum’s keynote reports focused on trends in industrial digitalization and intelligent transformation, sharing theoretical research results and practical applications of industrial vision, exploring the integration of machine vision with manufacturing, and promoting innovation and progress in the smart manufacturing industry.