Abstract: This research provides a new dataset, method, and evaluation system for automatic narration and understanding of PPT animations, promoting the application of multimodal vision-language models in dynamic presentations of structured documents. Future work will further expand the dataset, optimize inference efficiency, and explore the generalization capabilities of more animation types. This work is expected to lay the foundation for intelligent generation of slide animations, achieving an end-to-end generation process from static PPTs to dynamic narrated videos, providing transferable solutions for industries such as education, presentations, and digital publishing.

Speaker Bio: Changliang Xu is currently a researcher at the Hangzhou Institute for Advanced Study, University of Chinese Academy of Sciences, where he leads the Large Model Complex Agent Laboratory, focusing on automating complex workflows using multimodal large models and multi-agent systems. He also founded MuYu GongChang, dedicated to the research and commercialization of video production and knowledge construction agents. Previously, he served as General Manager of Big Data at Alibaba Cloud and CEO of Xinhua Zhiyun, a joint venture between Xinhua News Agency and Alibaba.

Abstract: In the era of real-time data-driven intelligence, visualization technology is becoming a core tool for information interaction and decision analysis. This talk will focus on the company's breakthroughs in core technology areas such as 3D graphics rendering engines, digital graphic asset cloud platforms, and real-time visualization toolchains, achieving dynamic integration and efficient rendering of massive multi-source data. By combining AI algorithms and interactive design, low-latency and high-precision visualization solutions have been created in fields such as smart cities, industrial simulation, and media communication. The talk will emphasize how to promote the paradigm shift from static display to intelligent interaction through collaborative innovation between real-time data platforms and visualization engines, providing reusable technical paths and practical cases for the industry.

Speaker Bio: Xingbo Tang, founder of EDP Technology Corporation, director of the China Association for Promoting Science and Technology Cooperation, vice chairman of the China Virtual Reality and Industry Technology Platform, senior expert in the Metaverse Innovation Exploration Matrix of the China Academy of Information and Communications Technology, and member of the China Society of Image and Graphics. Since 2003, he has led a team to develop domestic 3D graphics real-time rendering, tracking, recognition, processing engines, no-code creation tools, and digital asset management platforms. Relying on the digital content industry and combining cutting-edge technologies such as cloud computing, big data, artificial intelligence, and blockchain, he has built a closed-loop system for the entire digital content industry chain, widely applied in virtual reality, augmented reality, virtual simulation, digital twin, and industrial metaverse fields. He has participated in several national core technology research projects and topics.

Abstract: In the AI era, consumer products are experiencing a trend towards rich visualization. When business iteration speed exceeds UI production capacity, when user demand changes faster than design rhythm, and when data presentation requires personalized experiences, a revolution in UI/visualization production methods is occurring. We have launched the 🐯卧虎 - UI Generation Center, which enables 'data to interface in seconds', 'runtime UI generation solutions capable of handling hundreds/thousands/millions of QPS', 'multi-round dialogue', and 'creative enhancement', allowing frontend production to operate automatically and user experience to change on demand. This talk will share the implementation and prospects of new era visualization technology in large companies.

Speaker Bio: Zhaosong Huang, Alipay (Hangzhou) Information Technology Co., Ltd. Frontend/AIGC & Visualization Technology Expert. Obtained PhD in Computer Science from Zhejiang University. Mainly engaged in generative UI-related work.

Abstract: To address the three major pain points of traditional industrial robots, namely "weak scene generalization ability", "low human-robot collaboration efficiency", and "high cost of process iteration", this project relies on an embodied intelligent all-in-one machine. It is based on a self-developed embodied intelligent operating system (EOS) and a multimodal fusion model (ELM) that integrates language, vision, touch, and force perception. It covers the entire process toolchain (ETC) for task simulation, data collection, model training, inference deployment, and system integration to support various robot bodies and provide rich embodied robotic process packages. This will assist automation system integrators in quickly adopting robots driven by embodied intelligence in data visualization and virtual reality simulation applications.

Speaker Bio: Jinghui Wu, Chairman of Quanjing Technology Investment Group, currently serves as a distinguished researcher at the Suzhou Institute of Advanced Technology, Chinese Academy of Sciences, and is also the co-founder and Chief Scientist of Zhongke Shiyun. He is the director of Quanjing Future Laboratory and graduated with a Master's degree in MBA (TECH) from the University of New South Wales (UNSW) in Australia. He previously served as the Technical Director of IBM Australia's Watson Analytics, represented China at the BRICS Youth Summit on sustainable development themes related to big data, was elected Vice President of the Global Chinese New Generation Alliance, and serves as a distinguished expert for the Shenzhen Artificial Intelligence Association focusing on AI big data strategic digital transformation consulting and professional training. In 2015, he returned to China to start a data visualization analysis startup, which was acquired in 2021, becoming a technology investor and successfully investing in over 10 companies including Zhongke Shiyun, Shiyun Technology, Zhuhai Zhongke Shiyun Quanjing, Jijing Technology, Nanjing Quanjing, and Zhongke Chuangzhi through IT technology strategic investments.

Abstract: In the era of AI large models, the form of BI products has gradually evolved from traditional reports, dashboards, and self-service analysis to the new generation of Copilot, ChatBI, and Data Agent. How data visualization technology continues to add value in these new business forms and helps users improve data analysis and decision-making efficiency is a new topic in the field of data visualization. This talk will introduce AntV's exploration in the direction of data intelligent visualization technology and its practice in Ant Group's BI product DeepInsight.

Speaker Bio: Shaoting Deng, Frontend Technology Expert at Ant Group, Frontend Lead of BI Product DeepInsight, and Head of AntV AVA. Graduated from Wuhan University of Technology in 2017 with a Master's degree. Joined Ant Group in 2019 and currently leads the frontend of Ant Group's BI product DeepInsight, providing core technical support for the business through data visualization technologies such as AntV G2, S2, and AVA, serving over 70,000 internal users and commercializing benchmark data products externally. Technically focused on the field of data visualization analysis for nearly 5 years, has obtained 3 national patent authorizations, published 1 AntV visual solution white paper, and is currently focused on exploring and practicing AntV AVA data intelligent visualization.