Hong Kong Stock Concept Tracking | Apple Generates Digital Avatar in Half an Hour! AI Digital Humans Accelerate Implementation, Suggesting Focus on Various Segments of the Industry (with Concept Stocks)

Zhitong App has learned that Apple recently released a research paper discussing its ongoing research on generative AI technology called HUGS. According to reports, HUGS, which has been trained, can generate digital human avatars in about 30 minutes. It is said to be about 100 times faster than other methods such as NeuMan and Vid 2 Avatar. It is evident that Apple is making bold strides in innovation, and it is expected that new technological advancements will continue to emerge in related industries under the influence of artificial intelligence. Dongfang Securities suggests paying attention to various sub-sectors within the industry as AI-generated digital humans accelerate their implementation and related industry policies quickly catch up. Relevant concept stocks include SENSETIME-WR (00020), Baidu Group-SW (09888), and NetEase-S (09999).

Apple researcher Anurag Ranjan tweeted that HUGS stands for Human Gaussian Splats, which uses machine learning and computer vision to create realistic human elements with minimal input data. Apple's official introduction states that while neural rendering technology has made significant progress, it is still best suited for static scene photogrammetry and has not yet been extended to freely moving humans in an environment.

HUGS uses the 3DGaussian Splatting technique to create movable humans in a scene. This method requires a small amount of primary video, typically capturing the subject in motion within the scene and displaying as much surface as possible for the system to work. In some cases, the technology has low requirements for the source input data, with a minimum of 50 to 100 frames of monocular video, equivalent to 2 to 4 seconds of 24fps video.

In fact, Apple has been dedicated to research on digital avatars. For example, in its latest MR headset, VisionPro, the device can scan users' facial information through a front-facing camera and generate a digital avatar based on machine learning and encoding neural networks. When users make FaceTime calls, the digital avatar can mimic their facial and hand movements.

Furthermore, according to media reports, the biggest update for the iPhone 16 series is the introduction of AIGC. Apple plans to make significant upgrades to the microphones of the iPhone 16 to enhance the Siri experience with new artificial intelligence enhancements. In a recent post, Ming-Chi Kuo wrote that improving the hardware and software capabilities and specifications of Siri is crucial for promoting AI-generated content. He stated that Apple's ambitions in generative AI and the integration of large language models (LLMs) into Siri will heavily rely on improved speech input processing.

Industry insiders revealed that Apple plans to incorporate AIGC technology into iPhones and iPads as early as the end of 2024. In October of this year, when Apple released the M3, M3 Pro, and M3 Max chips for personal computers, it also revealed key information. Apple stated that these chips introduced an enhanced neural network engine for accelerating powerful machine learning (ML) models. In 2023, there is no doubt that it is a year of great momentum in the AI industry. The competition surrounding large models has sparked many industry trends. With the joint efforts of global technology companies, the application of AI has also moved from the cloud to the edge and end devices. Smartphones, PCs, and other consumer electronics are undergoing significant updates to accommodate the applications of large models and AIGC.

Looking ahead to 2024, the well-known technology media Tech Crunch has made six predictions for the AI industry trends next year: 1) OpenAI will become a product company; 2) Generative video and music production will become more mainstream; 3) Hybrid small models may have more advantages; 4) Many companies will abandon large-scale AI applications, leading to a wave of industry consolidation; 5) Apple is about to join the battle; 6) A completely new AI compliance industry may emerge.

Oriental Securities stated that AI empowerment enables digital humans to have a more intelligent core. The driving force behind virtual digital humans can be divided into "intermediaries" driven by real people and AI-driven. In the past, the pain points of AI-driven digital humans were high costs and poor interaction effects, so "intermediary" driven digital humans were more common. However, with recent technological breakthroughs in the AI field and the rapid implementation of large models in text, image, and other areas, the application scenarios and scope of digital humans will significantly expand. The accelerated implementation of AI digital humans and the rapid follow-up of related industry policies suggest paying attention to various sub-sectors of the industry.

In terms of investment opportunities, Cinda Securities recommends focusing on three major innovative areas in AI. Since 2023, the AI industry has made rapid progress and is currently at the forefront of global technological advancement. The bank believes that the three major innovative areas worth focusing on in the AI industry are large language models represented by GPT, AIGC represented by image/video generation technology, and hardware and autonomous driving represented by automatic control/edge computing. It is recommended to focus on technology giants such as Microsoft, Google, Baidu, Tencent Holdings, Alibaba, which have large language models, AIGC application leaders such as Adobe, Meitu, Fubo Group, and Nemo Film and Television, leading PC and smartphone brands in the edge hardware sector such as Lenovo Group and Xiaomi Group, and new forces in autonomous driving technology such as Xiaopeng Motors.

Related concept stocks:

SENSETIME-WR (00020): In September, SENSETIME-WR's "SenseAvatar" AI digital human generation platform passed the evaluation of the first batch of "Trusted Virtual Human Content Management System" by the China Academy of Information and Communications Technology, and was awarded the "L3 Excellence" certificate. This marks that SENSETIME-WR's SenseAvatar has reached the industry benchmark level in terms of functional completeness, system trustworthiness, and industrialization capabilities.

Baidu Group-SW (09888): VidPress, incubated by Baidu Research Institute, is the industry's first intelligent video synthesis platform that supports general and large-scale production. Users only need to input news text and image content links with a single click to automate the video production process, which can be completed in a few minutes. NetEase-S (09999): Currently, NetEase's self-developed AI technology has been applied to the entire process of game industrialization, with an efficiency improvement of up to 90% in key areas. It is reported that NetEase Interactive Entertainment AI LAB has independently developed dozens of AI productivity tools, covering various aspects such as AI voice generation, AI concept art generation, AI video motion capture, and AI model generation.