How does AI technology of artificial intelligence apply to the field of video?

As we all know, the era of national video has come, and users'attention has shifted from traditional text and pictures to video. It is believed that most users' mobile phones will have several on-demand, live and short video APPs. According to the Open Data Report, more than 70% of Internet traffic comes from video, which will exceed 90% in the future.

In the era of artificial intelligence, the key is how AI technology can truly play its application and commercial value in various industries and fields, and bring about industrial change. On March 28, Zou Juan, an expert in Aliyun video service technology, will lead you to explore how AI technology can be applied in familiar video scenarios to solve problems in actual business scenarios.

Throughout the whole video life cycle, including video capture, video production, video broadcasting and experience by the vast number of users. In fact, in this process, great changes have taken place in the entire video industry, from very professional participants to popular participation at every stage.

From the acquisition stage, professional video cameras and video recorders were first used for video acquisition, and now everyone is from the media producers, using mobile phones to shoot. In the production stage, from the heavyweight non-linear editing software to the special effects, filters and beauty of short video APP, basic video production can be carried out on the mobile phone. In the field of broadcasting, at first the TV station must have a program list. Now users can play on-demand video on the website and use mobile phones to achieve personalized search and viewing. In the experience of this link, users from the most original cable coaxial cable single transmission viewing, now we can interact, comment, praise, bullet screen and so on. Therefore, we can see that, from the whole video life cycle, participants have undergone tremendous changes, and more and more applications are joining the video field.

As can be seen from the figure below, the combination of video and AI has run through every stage of the video life cycle.

So what can AI bring to the video industry?

First, improve production efficiency

The combination of AI and acquisition and production can effectively improve the efficiency of video production. Traditional editing is done by human. When AI is combined with video capture and production, we can introduce intelligent editing technology to produce video quickly. The world's martial arts, only fast and unbreakable, applied in the field of video is the same. Imagine that we can produce videos very quickly and publish them online at the first time, so that we can get more traffic.

Second, avoid regulatory risks

Introducing artificial intelligence auditing technology into video can shorten the video publishing cycle, reduce the interference of manual auditing, and more efficiently and accurately avoid regulatory risks.

Third, release manpower and reduce costs

Because in the first two stages, machines and algorithms do a lot of human replacement operations, so the release of manpower, and can bring about cost reduction.

Based on the AI algorithm of Damo Institute, combined with the accumulation of video cloud team in the field of audio and video technology for many years, Aliyun constructed the video AI capability - retina, and divided the product function into four modules: auditing, recognition, understanding and search. The capabilities shown below are actually the basic capabilities of the smallest unit of the video cloud AI service. In fact, they can be combined based on these capabilities. Like building blocks, they can penetrate into various video scenarios to form solutions for various matching services.

Next, Mr. Zou Juan introduced three scenarios in the field of video production: acquisition, production, distribution and broadcasting, and media management. In these three scenarios, Aliyun grew up with customers, penetrated into the actual business scenarios of customers, and combined with its own product planning, introduced the solution of video AI.

In the scene of acquisition and production, the solution supports video shooting, caption, editing synthesis and video splitting; in the scene of video broadcasting, besides common video auditing, copyright detection is gradually recognized by the public, as well as target recognition and detection in the process of real-time broadcasting; in the scene of media management, the most classic is intelligent cataloguing and intelligent cover. The solution also includes fingerprint-based weight removal and the ability to mine and organize between videos.

We can see that through the combination of basic AI capabilities and the actual business scenarios of customers, Aliyun has provided some specific solutions. Here are some typical scenarios to introduce the specific landing.

First, video capture scene - video shooting

In the popular short video and photo APP in the market, sticker function based on face recognition technology is basically provided. Many customers hope to have a rich shooting effect. Aliyun provides real-time video processing function on the solution of short video intelligent terminal. It has built-in face recognition and dynamic sticker library. In the future, it can also pay to upgrade advanced functions such as big eyes, thin face and so on.

Second, Video Production Scene - A Brilliant Collection

This is a business scenario that many customers have and there are many AI technologies available. The AI technology on the left of the figure below, combined with cloud video editing capabilities, can do a lot of imaginative things. For example, the time line of the appearance of characters is linked together to automatically generate a collection of characters; for example, the classic sports events are brilliant moments, the pre-material sorting can be automatically processed through AI; for example, the application of thematic production in this field is also very extensive, such as last year's snow in Jiangnan, radio and television.

Please read the Chinese version for details.