WiMi Hologram Cloud Inc (WIMI)

[StockpickerCAPC]

Facebook's new AI model realizes picture monitoring, WiMi holographic light field vision AI modular splicing algorithm is excellent.

source
https://finance.sina.com.cn/stock/relnews/us/2021-03-19/doc-ikknscsi8664099.shtml

March 19, 2021 15:27 Tencent Watchlist

With the breakthrough of deep learning technology, artificial intelligence has begun to develop rapidly on a global scale at an unprecedented speed. China's artificial intelligence technology and practical applications have walked in the forefront of the world, and have achieved results in various fields of industry. That is, the AI ??algorithm is the code, a new paradigm of AI programming based on the native expression of mathematics, to create an easy-to-use, efficient, and easy-to-debug micro-programming architecture, lower the development threshold, and allow algorithm experts to focus on AI innovation and exploration.

At present, computer vision is one of the most popular research fields in deep learning. It is located at the intersection of many academic subjects, such as computer science (graphics, algorithms, theory, systems, architecture), mathematics (information retrieval, machine learning), engineering, robots, voice, Natural language processing, image processing), physics (optics), biology (neuroscience) and psychology (cognitive science). Because computer vision represents a relative understanding of the visual environment and its background, many scientists believe that this field paves the way for artificial intelligence because of its cross-domain mastery.

Currently AI computer vision applications are everywhere, including autonomous vehicles, smart phones, surveillance cameras, consumer cameras, AR/VR, robotics and industrial applications, etc. Although all walks of life are accelerating the layout and development of artificial intelligence products, possessing the core technology of artificial intelligence is the key to achieving high-quality product application development.

Obtaining training data is one of the biggest competitive advantages in the field of artificial intelligence. By collecting this resource of millions and tens of millions of users, technology giants such as Facebook, Google and Amazon have been able to achieve leading advantages in various fields.

The company wrote in its official blog:
"By learning global public video streams spanning hundreds of languages ??across multiple countries, our artificial intelligence system will not only improve accuracy, but will also adapt to our fast-moving world and recognize The nuances and visual cues of different cultures and regions".

Facebook said that the resulting machine learning model will be used to create new content recommendation systems and control tools, but more can be done in the future. Of course, Facebook can already obtain such information through the current advertising targeting business, but if it can parse the video through artificial intelligence, it will add an incredibly rich (and intrusive) data source to its store.

Recently, Facebook shared the details of two internal artificial intelligence projects, namely Learning from video and TimeSformer. These two projects aim to promote the development of more powerful machine learning models.

One of the first projects, "Learning from video" (Learning from video), Facebook will use the videos uploaded by users to train the machine learning system that drives its social network.

Facebook is using a "self-supervised learning" method to better use user video training models, this method does not need to tag the training data.

The model is called SEER (SElf-SupERvised) and is fed with 1 billion publicly available Instagram images that have not been manually tagged. However, even without the labels and annotations that are usually used in AI algorithm training, seeder can still train the data set autonomously, continue to learn, and finally achieve the highest accuracy on tasks such as object detection.

This method, called self-supervised learning, is already very mature in the AI ??field: it consists of systems that can learn directly from given information, without having to rely on carefully labeled data sets to teach them how Perform tasks such as recognizing objects in photos or translating text.

Self-supervised learning has attracted a lot of attention recently, because it means that there is less work that requires manual labeling of data, which is time-consuming and laborious work for most researchers. Without the need to manage data sets, the self-supervised model can handle larger and more diverse data sets.

In some fields, especially natural language processing, this method has made breakthroughs. Training algorithms on an increasing amount of unlabeled text has led to progress in applications such as question answering, machine translation, and natural language inference.

At present, artificial intelligence, as a strategic technology leading the future, has become a new high ground for global competition. Artificial intelligence has become the primary focus of my country's scientific and technological development. To speed up its development process, it is necessary to obtain sufficient AI computing power support behind it.

In recent years, China's investment in artificial intelligence has shown a rapid growth trend.Led by ultra-large-scale Internet companies and large industry users, companies have increased their investment in artificial intelligence to meet the increasing demand for computing power in AI innovation. And seek for the digital transformation of enterprises and the overall industrial transformation and upgrading of China.

WiMi Hologram represented by computer AI vision, WiMi's computer vision holographic cloud service, and its commercial application scenarios are mainly concentrated in five professional fields such as home entertainment, light field theater, performing arts system, commercial publishing system and advertising display system, and have been deployed to Corresponding areas of smart cities. WIMI focuses on holographic cloud services, mainly focusing on automotive AR holographic HUD, 3D holographic pulse LiDAR, head-mounted light field holographic equipment, holographic semiconductor, holographic cloud software, holographic car navigation and other professional fields, covering from holographic car navigation AR technology, 3D holographic pulse LiDAR technology, holographic vision semiconductor technology, holographic software development, holographic AR advertising technology, holographic AR entertainment technology, holographic ARSDK payment, interactive holographic communication and other holographic AR technologies. It is a holographic cloud integrated technology Solution provider.

WiMi Hologram has the mission of "the horizon is the horizon". The company has established the world's top, self-developed deep learning platform and supercomputing center,
and has developed a series of AI technologies, including:
Face recognition, image recognition, text recognition, medical care Image recognition, video analysis, unmanned driving and remote sensing, etc. The development of holographic 3D face recognition software is based on WiMi's holographic imaging feature imaging detection and recognition technology, template matching holographic imaging detection technology, and video processing and recognition technology based on deep learning and training. Traditional 2D facial recognition technology is a recognition technology based on facial features, which captures information from facial images or facial video streams, and automatically detects and tracks the target face; WiMi's holographic 3D facial recognition technology is holographic imaging capture and 3D portrait The combined recognition technology.

The holographic face change technology is based on the holographic 3D layer replacement technology, including AI-based image recognition and dynamic fusion processing technology, real-time tracking of images and replacing faces with other faces. This technology replaces the human face in the video frame, synthesizes the video and adds the original audio. WiMi has verified these technical modules in the holographic AR plug-in advertising application, and will continue to develop and upgrade these technical modules. WiMi believes that this technology will bring new business growth to applications such as celebrity advertising, movie distribution and live video streaming.

WiMi’s leading holographic AR content production function is built around image acquisition, object recognition, automatic image processing and computer vision technology. WiMi’s software engineering team and visualization design team work closely to advance these visualization-related technologies and use them to design and produce innovative holographic AR content. Through real-time computer vision algorithms that provide accurate pose estimation, scene recognition and tracking can be performed in a few seconds. This cutting-edge algorithm also allows WiMi to perform photorealistic high-resolution rendering visualizations on a pixel-based basis. Frost & Sullivan said that although most peer companies may identify and capture 40 to 50 pieces of image data in a specific space unit, the number of data pieces that WiMi can collect reaches 500 to 550; WiMi’s image processing speed is 80% higher than the industry average. %, thereby improving operational efficiency. In the scene reconstruction process, WiMi’s automatic image processing tools can perform noise removal and feature enhancement on the originally captured image, thereby creating a best-in-class holographic AR design with industry-leading simulation.

In 2020, China's artificial intelligence server will account for about one-third of the global artificial intelligence server market, becoming the backbone of the global artificial intelligence industry.

In 2020, facing the ever-increasing wave of computing power demand, we will forwardly propose the "intelligent computing center", through the new public computing power infrastructure to provide computing power services, data services and algorithm services required for artificial intelligence applications to carry AI technology innovation. Promote the open sharing of data, accelerate the construction of smart ecology, and drive the convergence of smart industries.

Recently, International Data Corporation released the 2020H1 ``Global Artificial Intelligence Market Semi-Annual Tracking Report''.The report shows that the global artificial intelligence server market reached 5.59 billion U.S. dollars in the first half of 2020, accounting for more than 84.2% of the artificial intelligence infrastructure market, becoming an AI infrastructure The main body of demand.

In the past year, the application of artificial intelligence in the industry has developed rapidly, and general-purpose application scenarios have achieved considerable maturity. Driven by business needs, fragmented applications with high industry attributes have also begun to be widely used and radiated to Media and entertainment, modern agriculture, smart home, smart power and many other fields.

At present, the computing power required for artificial intelligence doubles every two months, and the supply level of new computing power infrastructure that carries AI will directly affect AI innovation iterations and industrial AI applications.