China AI Native Industry Insights – 20241107 – Tencent | Alibaba | ByteDance | XPENG Motors | more
1.Tencent Hunyuan Open Source Hunyuan3D-1.0: The first 3D open-source large model that supports both text-to-image and image-to-text generation.
Tencent has unveiled Hunyuan3D-1.0, a groundbreaking open-source 3D model that is the first of its kind to support both text-to-image and image-to-3D generation. This dual-phase model can create 3D assets in just 10 seconds while maintaining high quality, showcasing impressive versatility across various object scales. It marks a significant advancement in 3D modeling technology, enhancing accessibility for developers and creators.
Original Chinese article: https://www.ithome.com/0/808/138.htm
English translation via free online service: https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=https%3A%2F%2Fwww.ithome.com%2F0%2F808%2F138.htm
2.Alibaba Damo Academy releases the Eight-View Meteorological Model: Key indicator prediction performance surpasses traditional weather forecasts.
Alibaba Damo Academy has unveiled an advanced weather modeling system, aptly named the Eight-View Meteorological Model, which significantly outperforms traditional forecasts in key predictive metrics, including wind speed and temperature. This innovative model integrates regional data to enhance spatial and temporal resolution, achieving remarkable prediction accuracies of over 96% for renewable energy output and 98% for power load, thus advancing decision-making in various fields such as agriculture, aviation, and sports.
Original Chinese article: https://www.163.com/dy/article/JGB28L9V0511CPVM.html
English translation via free online service: https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=https%3A%2F%2Fwww.163.com%2Fdy%2Farticle%2FJGB28L9V0511CPVM.html
3.ByteDance has launched the X-Portrait2 model, which enables one-click generation of identical facial expressions.
ByteDance has unveiled its advanced video-driven model, X-Portrait 2, which allows users to generate high-quality, expressive videos from a single static image and a driving video. This innovative technology streamlines the creative process by preserving the original identity while capturing both subtle and exaggerated emotional expressions, enhancing movement representation significantly compared to previous methods.
Original Chinese article: https://www.ithome.com/0/808/457.htm
English translation via free online service: https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=https%3A%2F%2Fwww.ithome.com%2F0%2F808%2F457.htm
4.Significantly enhance SAM 2 without training! The open-source SAM2Long is here, produced by the Chinese University of Hong Kong and Shanghai AI Laboratory.
A research team from Chinese University of Hong Kong and Shanghai AI Laboratory has introduced SAM2Long, an innovative extension of Segment Anything Model 2 (SAM 2), addressing challenges in long video object segmentation. By implementing a multi-path memory tree structure, SAM2Long improves robustness and accuracy, minimizing the impact of errors in segmentation masks across frames. This advancement significantly outperforms its predecessor, SAM 2, across various datasets, promising broad applications in areas like autonomous driving and video editing.
Original Chinese article: https://www.jiqizhixin.com/articles/2024-11-05-5?from=synced&keyword=%E5%BC%80%E6%BA%90%E7%9A%84%20SAM2Long%20%E6%9D%A5%E4%BA%86%EF%BC%8C%E6%B8%AF%E4%B8%AD%E6%96%87%E3%80%81%E4%B8%8A%E6%B5%B7%20AI%20Lab%20%E5%87%BA%E5%93%81
English translation via free online service: https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=https%3A%2F%2Fwww.jiqizhixin.com%2Farticles%2F2024-11-05-5%3Ffrom%3Dsynced%26keyword%3D%25E5%25BC%2580%25E6%25BA%2590%25E7%259A%2584%2520SAM2Long%2520%25E6%259D%25A5%25E4%25BA%2586%25EF%25BC%258C%25E6%25B8%25AF%25E4%25B8%25AD%25E6%2596%2587%25E3%2580%2581%25E4%25B8%258A%25E6%25B5%25B7%2520AI%2520Lab%2520%25E5%2587%25BA%25E5%2593%2581
5.Alibaba tests AI animation creation tool “Animode”: supports one-click animation of videos.
Alibaba is testing “Animode,” an AI-driven video creation tool designed to streamline the anime production process. Users can effortlessly transform their videos into fluid two-dimensional styles by simply uploading their materials. With advanced features like real-time motion capture and built-in action libraries, Animode enhances the creative experience, making it easier for creators to produce high-quality content.
Original Chinese article: https://www.aibase.com/zh/news/13054
English translation via free online service: https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=https%3A%2F%2Fwww.aibase.com%2Fzh%2Fnews%2F13054
6.ByteDance enters the AI video generation field with the internal testing of its AI assistant Doubao for video generation.
ByteDance has officially launched the beta testing of its AI assistant Doubao’s video generation capabilities, entering the competitive AI video creation market. The new model allows for seamless transformation of images and text into videos, featuring dynamic cinematography and various style options suitable for sectors like e-commerce and education. With advanced understanding of semantics and complex interactions, Doubao promises a significant enhancement in video generation technology.
Original Chinese article: https://www.ithome.com/0/808/600.htm
English translation via free online service: https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=https%3A%2F%2Fwww.ithome.com%2F0%2F808%2F600.htm
7.XPENG Motors has released its “Turing” AI driving system, aiming to achieve one takeover per 100 kilometers by next year.
At the recent XPENG AI Tech Day, XPENG Motors unveiled its “Turing” AI driving system, highlighting its fully self-developed cloud, software, and hardware stack. The system features a cloud model with 80 times the parameters of its vehicle counterpart, and XPENG’s CEO announced plans for a powerful cloud computing capability of 10 Eflops by 2025, aiming for an advanced autonomous driving experience akin to learning from a Nobel-level instructor. The introduction of the Turing AI chip and the “Canghai” operating system further enhances vehicle perception and responsiveness for L4 autonomous driving.
Original Chinese article: https://www.ithome.com/0/808/370.htm
English translation via free online service: https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=https%3A%2F%2Fwww.ithome.com%2F0%2F808%2F370.htm
8.XPENG Iron, the XPENG AI robot, makes its debut featuring an end-to-end AI Eagle Eye vision system.
At the 2024 XPENG AI Tech Day, the new XPENG AI robot, Iron, was unveiled. Standing 178cm tall and weighing 70kg, Iron boasts 62 degrees of freedom and incorporates the advanced XPENG Eagle Eye vision system, enabling autonomous and natural bipedal movement. Currently, Iron has commenced work in XPENG’s factory, assembling components for the upcoming P7+ model.
Original Chinese article: https://www.ithome.com/0/808/383.htm
English translation via free online service: https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=https%3A%2F%2Fwww.ithome.com%2F0%2F808%2F383.htm
9.Autonomous driving AI company DeepRoute.ai completed a $100 million C1 round of strategic financing.
DeepRoute.ai has successfully secured $100 million in Series C1 financing, solely backed by leading domestic automaker Great Wall Motors. This funding will bolster their mass production projects, expand international operations, and support the exploration of Robotaxi commercialization and cutting-edge VLA model technologies. CEO Zhou Guang emphasized the transition from electric vehicles to the competitive arena of intelligent driving over the next five years.
Original Chinese article: https://www.aibase.com/zh/news/13069
English translation via free online service: https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=https%3A%2F%2Fwww.aibase.com%2Fzh%2Fnews%2F13069
10.Jinke Tom Culture announced that the main functions of the AI robot and the AI storytelling application have been successfully developed.
Jinke Tom Culture is diving into AI development, revealing plans to launch AI robots and storytelling applications. The company has completed the primary development for these products and is now moving forward with testing and deployment, leveraging its unique digital content expertise for a more interactive parental experience.
Original Chinese article: https://ai.zol.com.cn/916/9160001.html
English translation via free online service: https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=https%3A%2F%2Fai.zol.com.cn%2F916%2F9160001.html
That’s all for today’s China AI Native Industry Insights. Join us at AI Native Foundation Membership Dashboard for the latest insights on AI Native, or follow our linkedin account at AI Native Foundation and our twitter account at AINativeF.