Yuelai Guangdong AI series report
This year's government work report proposes to vigorously promote the construction of a modern industrial system and accelerate the development of new quality productive forces. It mentions that it is necessary to deepen the R&D and application of big data and artificial intelligence, carry out the "artificial intelligence +" action, and build a digital industrial cluster with international competitiveness.
Soon after the release of the "22 Articles for the Development of General Artificial Intelligence Industry", Guangdong made a New Year's deployment of "accelerating the cultivation of emerging industries such as artificial intelligence and vigorously developing new quality productivity". This year, the Nandu Big Data Research Institute will carry out a large-scale research on "Guangdong to Guangdong AI", which will describe the development status in multiple chapters, consider the application effectiveness, study and judge the governance risks, help Guangdong accelerate the construction of a leading place for general artificial intelligence industry innovation, and collaboratively explore the innovation of artificial intelligence regulatory models. This time, we launched a series of reports on "AI Applications", visited artificial intelligence-related enterprises in Guangdong, and revealed the research and development stories of AI application products that are closely related to our work and life.
In the second phase, we walked into Shenzhen Qianhai Hand-painted Technology and Culture Co., Ltd. to see how AI gives a soul to digital photo frames and becomes a hit in the Middle East market.
Unexpectedly, a talking AI digital photo frame has become a hit in the Middle East. Just upload a portrait photo and a sound clip to generate a digital human in the frame, he/she can not only clone your voice, but also speak more than 50 languages and communicate with users without barriers.
This photo frame comes from an AIGC brand in Shenzhen, and can be used in scenes such as exhibition hall tours, family companionship, and mourning for the deceased. Since the beginning of this year, the brand has appeared in many exhibitions at home and abroad, such as the Guangdong Provincial High-quality Development Conference Industrial Science and Technology Integration Development Achievement Exhibition, the 2024 Saudi Arabia Communication Technology and Information Technology Exhibition, etc.
There is an AI product: AI photo frames
Product Features:
With a single photo, you can create a video of a digital human that can be used to have a conversation.
Application Scenarios:
It can be used as a service assistant in government affairs/finance/medical and other scenarios, as well as an explanation assistant in museums, cultural centers, exhibition halls, etc., to accompany the empty nest elderly, mourn the deceased, and convey condolences.
Effects:
Enhance the interest of explanations, provide users with relevant information and personalized services, and interact with the audience in real time around the clock and in an all-round way.
Launching the "first shot" in the Middle East
"The AI photo frame is inspired by Dubai's golden image frame, with the generation technology of AI digital human, you can talk to anyone in the world, this idea reproduces the talking photo frame in the Harry Potter movies", Wei Bo, the founder of Laihua, introduced that Middle Eastern countries have a cultural custom of hanging photos of distinguished people on the wall, and this market potential was keenly captured by him who came to Saudi Arabia to expand his business, and the AI photo frame was born.
On February 18th, Laihua AI photo frame was unveiled at the Guangdong Provincial Industrial Science and Technology Integration Development Achievement Exhibition.
Positioned as "good-looking skin, interesting soul", but the AI photo frame could only achieve simple special effects at first, "the skin is not intelligent". In Weibo's view, the birth of ChatGPT has truly given the soul to the AI photo frame, which can be used to dialogue with the characters in the photo frame and generate videos by binding the APP and uploading the corresponding materials, which can currently support more than 50 languages. In foreign countries, some football clubs put the AI digital human photo frames of stars such as Neymar and Ronaldo into fan stores for sale, and they were "crazy robbed".
Laihua's co-founder Quan Qianhui told a reporter from Nandu that the AI photo frame is the company's "first shot" in the Middle East, and she was also surprised by the cultural and emotional sustenance of the Arabs in the entire Middle East for the "photo frame". It is understood that this AI digital photo frame has been sold overseas tens of thousands of units, not only hung in the Abu Dhabi Royal Palace, but also as a national gift by the royal family to distinguished guests from afar.
From photo frames to AI assistants
"It's more like the role of an AI assistant," said Quan in an interview, preferring to define AI photo frames as technology-driven consumer electronics. By integrating the Al Assistant SDK into different hardware, it can realize the connection of multi-industry intelligent interaction scenarios.
Not only can it be made into a "photo frame", but the AI assistant can also be incarnated as a café orderer, museum docent, corporate front desk, etc., to answer questions in government affairs, finance, medical and other scenarios, carry out popular science education in museums, cultural centers, exhibition halls and other scenes, and become a spiritual companion for the empty nesters, a medium to mourn their deceased relatives, and so on. Nandu reporters noticed that the AI assistant has been applied to the IP assistant of the COP28 United Nations Climate Change Conference, the AI guide assistant of the Abu Dhabi Grand Mosque, the AI assistant of the official website of the Abu Dhabi Police Department, and the Abu Dhabi café ordering assistant.
Quan Qianhui said that in the field of culture, AI assistants can also be applied to cross-time and space dialogue scenes with Confucius, Lu Xun, Mona Lisa, Einstein and other ancient and modern Chinese and foreign celebrities, such as "Thirty Thousand Miles of Chang'an" Li Bai's animated image appeared on the stage of this year's Spring Festival Gala, and Laihua also took the opportunity to launch an AI assistant with Li Bai's image, users can talk with him and have interesting interactions.
Let's draw a new product: the AI assistant of Li Bai's image.
How AI creation is made
In January this year, Laihua released its 2023 annual report, and AI animation, AI digital human video, AI assistant, and AI photo frame have become the core keywords of the product.
At first, Laihua's positioning was not AIGC, but the idea of "making video creation easier". In 2015, Shenzhen Qianhai Hand-Painted Technology and Culture Co., Ltd., to which Laihua belongs, was founded, "We did not respect AIGC in the early days, and the reason why we have developed into an AI company step by step is that we continue to implant small AI functions into our products to empower users." Quan Qianhui revealed that the early idea of the company's establishment was to create animation technology, study the automatic hand-drawn path of animation, continuously iterate the user experience based on the animation intelligent generation platform, and implant functions such as custom IP cartoon images into the product, so that users can create characters and drive their speech, which is also an early digital human technology.
With the digital human out of the circle, Laihua Technology began to hire artists on a large scale to hand-draw a digital human IP template. According to Quan Qianhui, the driving technology of Laihua can be used to remotely control the image of a digital human by uploading a photo and connecting it to the mobile phone APP, so as to generate digital oral video, digital human animation, digital human live broadcast and so on. "Today's Laihua Technology has transformed into an AI creation platform. Quan Qianhui said.
30 million+ data-driven models
In 2023, artificial intelligence will usher in a "singularity": the large language model ChatGPT will be born, and "generative AI" will form a surging tide, and new generative AI products will be born almost every day. Taking advantage of the east wind, Laihua relies on its own 18 million animation data assets to launch SkinSoul, a vertical model in the animation field, which has become the driving force of the platform's intelligent content creation and interactive experience capabilities, thereby supporting the creation of video and animation.
FOR EXAMPLE, USERS CAN USE URL TO VIDEO, A COMMERCIAL VIDEO PRODUCTION TOOL LAUNCHED BY THE PLATFORM, AND IMPORT LINKS TO AMAZON, TAOBAO AND OTHER WEB PAGES, SO THAT THEY CAN QUICKLY GENERATE A VARIETY OF VIDEOS IN BATCHES, AND CAN PERFORM SIMPLE DRAG AND DROP ON THE AL ANIMATION ONLINE CREATION PLATFORM TO GENERATE ANIMATED VIDEOS. At present, this function has been used in online education, media publicity, advertising and marketing, finance, medical and other industries, such as helping teachers convert course content into relevant videos with one click, quickly producing vivid courseware, and then using digital humans to explain to improve the fun and efficiency of teaching;
In more than 8 years, Laihua Technology has accumulated 30 million+ animation data assets, which have become the raw materials for training models. Quan pointed out that in terms of large model research and development, Laihua selects and trains the application layer based on open source large models, and tests the effects generated by different large models.
At present, Laihua has cooperated with the large model startup Moon Dark Side, based on the large model Kimi to generate and train "controllable videos", so as to achieve a strong match between content and picture. It is reported that Kimi excels in long text processing capabilities and can quickly extract key information from complex and lengthy documents. Kimi's smart assistant supports the input of 200,000 Chinese characters, which is about 400,000 tokens (the smallest unit of meaning that the model can understand and generate) according to OpenAI's calculation standard, which is the longest context input length that can be supported in the world's large model products at that time.
"The products produced, whether it is the background materials used in them or the digital people used, they are all genuine and commercial," Quan stressed Hui.
Go overseas to expand the global territory
For Laihua, going abroad with an "AI photo frame" has become a major turning point in the development of the brand. Many wealthy countries in the Middle East, represented by Saudi Arabia, the United Arab Emirates, Qatar, Kuwait, etc., have deployed various "2030 visions" to transform from energy-based countries to diversified and digital countries.
Since 2022, the company has begun to focus on the Middle East market, and has taken the UAE as the first stop to set up the Abu Dhabi Middle East headquarters to expand its global business territory. Its overseas animation product Doratoon and overseas digital human product InnAIO have attracted wide attention, and have also brought tens of millions of users covering more than 100 countries and regions around the world. Since last year, Laihua's products have been unveiled at well-known overseas events such as the Turkish National Summit, GITEX GLOBAL 2023 Global Science and Technology Event, and Saudi Arabia LEAP Science and Technology Exhibition, and have reached AI digital human cooperation with many institutions and enterprises such as Dubai China-Arab Satellite TV.
"When participating in these exhibitions, our AI photo frames are often 'crazy to be sold out', because of different cultural attributes, there will be different needs, forming unexpected 'explosive products'." When it comes to the next step of product planning, Quan said that when expanding overseas markets, she will recognize the overseas AI industry trends earlier, and will also bring this information to China in a timely manner to empower the next product research and development direction.
Producer: Nandu Big Data Research Institute
Curator: Zou Ying
Co-ordinator: Zhang Chun
Written by: Nandu reporter Zhang Yuting and Li Weifeng
Design: Zhang Bo