[Reading] Computer vision technology makes video intelligence constantly upgraded, Internet applications work in video, and make video a super entry point for Internet applications.
Artificial intelligence is undoubtedly the hottest keyword in the field of Internet technology in 2017. From all aspects of performance, artificial intelligence has entered a new turning point in 2017, which largely indicates that the commercial dividend period of artificial intelligence is coming, artificial intelligence is Applications in various fields have begun to break out.
In this context, computer vision, one of the three major areas of artificial intelligence (AI), has recently attracted more and more attention. As a basic technology application of artificial intelligence, computer vision has diverse usage scenarios and huge market potential. Therefore, under the traction of artificial intelligence, the computer vision-enabled video industry has also entered an era of intelligence.
Before entering the intelligence industry, the video industry has undergone three stages of changes under the influence of the Internet: first, the first stage is the traditional video era, and it is still in the stage of offline video and film as video transmission. I couldn't interact, and I couldn't choose the content myself. Later, I entered the Internet era and began to pop up online video platforms like Youku, Tudou, and Iqiyi. Users can choose to watch content on the platform through the network. This is the second stage; Until the rapid outbreak of live broadcasting in 2016, the video industry entered the third stage, that is, the C2C communication era. The platforms like Betta, YY, Yingke, and Second Shot are characterized by fast, accurate and fragmented propagation. The technical background is the link to H5, so users can send it at will.
Then the next step is to enter the current era of intelligence, in the face of face recognition, deep learning, etc., which are like the hot words of the industry, the emergence of technology-oriented new companies, computer vision technology upgrade video industry, let video in security , advertising marketing, new retail and other areas have been fully upgraded, so we come to inventory today, what computer vision can actually be applied to the video scene technology?
1. One of the application points: security field
The security field has become the first landing point for the combination of [computer vision + video]. This is mainly due to the two characteristics of security itself: First, the security industry with video technology as the core has a large number of data sources, which can fully meet the requirements of artificial intelligence for algorithm model training. Second, the pre-existence prevention and response in the security industry The appeal of tracing afterwards is in complete agreement with the technical logic of artificial intelligence.
At present, the application of [Computer Vision + Video] in the field of security mainly involves the recognition of faces and vehicles, including biometrics, big data and video structuring. Among them, biometric identification includes fingerprint recognition, iris recognition, face recognition, gait recognition, etc. The first two are mainly used for identity authentication in specific scenes; while on video structuring technology, it mainly integrates machine vision. Artificial intelligence technologies such as image processing, pattern recognition, and deep learning are also the basis for understanding video content. And there are application scenarios in many fields such as public security, transportation, building, finance, industry, and civil.
For example, the needs of users in the public security industry are in the vast amount of video information, the clues of criminal suspects are found. To achieve this, it is not enough to just capture the suspect with a camera. It needs intelligent front-end camera to analyze video content in real time, detect moving objects, and identify attribute information such as people and vehicles. Then, it needs to aggregate massive city-level information to the central database of back-end artificial intelligence for storage, and then use computing power and intelligent analysis capabilities. Real-time analysis of the suspect's information, and finally give the most likely clues.
Judging from the current market situation, the huge market size and considerable revenue profit prospects in the security field have just made it a must for many giants and startup companies. The traditional giants are led by Hikvision and Dahua. Actively lay out upstream key technology areas such as chips and algorithms. On the other hand, it is also extending the integration of integrators or operators to the downstream.
In the computer vision, the entrepreneurial unicorn company has completed the $410 million Series B round of financing in July this year, and the company that completed the billion-dollar C-round financing at the end of 16 years. It has the advantage of technical algorithms but it is difficult to achieve commercialization independently. In the early stage, the market layout was achieved through cooperation with traditional giants.
2. Application point 2: new retail
What is new retail, Ma Yun's definition is to use the advanced technology of big data, artificial intelligence and other means to upgrade the production, circulation and sales process of goods, and then reshape the structure and ecosystem, and serve online. , offline experience and a new retail model of deep integration of modern logistics.
Among the various solutions, the most widely mentioned is the visual recognition technology based on convolutional neural networks. How does computer vision combine with video in the new retail field?
That is, through deep learning, the church computer knows the goods. When the consumer self-purchases, the computer recognizes the category price of the goods through the in-store camera, and the consumption can be automatically deducted without the consumer's manual checkout.
Computer Vision gives video powerful recognition technology, launches a complete solution for new retail, integrates face recognition analysis engine, verifies user identity, and guides new users to payment method binding and identity/credit information entry. At the same time, combined with the face attribute analysis technology, the user's gender, age and other information are determined to generate a user portrait. Tracking the user's walking route, combined with face recognition technology, collects the length of time the user stays in front of the shelf, identifies the behaviors and products that the user takes and puts back, and analyzes the shopping trends and preferences of different users. Through the analysis of a large number of user image behavior data, the merchant is provided with suggestions for placing the merchandise shelf layout.
For example, Amazon launched the new concept store Amazon Go. Through high-tech such as AI and deep learning, customers only need to download the Amazon Go app. After scanning the code at the store entrance, they can enter the store and start shopping. Amazon Go's sensors calculate the customer's effective shopping behavior and automatically charge the Amazon account based on the customer's spending after the customer leaves the store. It involves computer vision, sensors, deep learning and other technologies. The core is to identify actions, goods and people, and to associate them by position or posture. Amazon also calls it "just walk out" technology.
3, application point three: video marketing
With the development and changes of the video industry, especially the rapid growth of mobile video, the way of Internet video advertising and marketing has also changed. At the same time, advertising revenue is still the mainstay of online video industry revenue. The gradual upgrading of video advertising marketing, advertisers' decision-making is no longer just a simple implant of water and floating on the surface. Then video sites, no matter from the user base or growth rate, no doubt have a huge imagination.
In this context, the industry chain-related enterprises began to focus on providing advertisers with more content marketing forms, higher advertising efficiency and more accurate advertising strategies, and the commercial value of video was further explored. In this context, the industry chain-related enterprises began to focus on providing advertisers with more content marketing forms, higher advertising efficiency and more accurate advertising strategies, and the commercial value of video was further explored. Among them, Video++ is based on artificial intelligence intelligent algorithm, which has the advertising value of video content, and creates a creative interactive form within the video to become the new favorite in the industry.
The consumer-grade video created by Video++ is a strange concept for most people. Consumer-grade video can refer to live video, movies, variety shows, TV series and other videos with complex scenes and large-scale processing in the late video category. With the development of the Internet, the data of many products are transmitted and stored through the Internet and the cloud. The large amount of data also makes the commercial value of consumer-grade video huge. The video itself is a blue ocean of advertisements. There are many commercial monetization methods. Without disturbing the user experience, adding some entertainment interaction methods allows users to complete a kind of e-commerce or advertising commercial value conversion in participation.
For the video, the content layer is the most special. In terms of content, AI can let the machine owner think, and when the video is transmitted to the machine for identification and analysis, it can identify stars, objects, brands, mobile phones, scenes, etc. Make the machine understand the content of the video as humans and discover interesting points. At the logical and application layers, these points can be commercialized using core components and video applications, and the structured data identified by the machine is applied as a delivery point to the advertising and e-commerce scenarios.
This year, Video++ has deep cooperation with Sohu. Video++ provides Sohu with AI identification technology services in the consumer video field, a complete video structured data application system, and an overall video AI data application solution based on video AI data applications. Meta-information screening, value weight function retrieval, etc., Sohu and Video+ have deep cooperation on content marketing products, providing content marketing solutions for advertisers, so that Sohu's various exciting content can be fully exploited.
When computer vision encounters the same high-demand video advertising marketing, technology and creativity are perfectly combined to connect brands and users, form multi-style interactive advertising, scene marketing, with AI's empowerment, the “vision†world will be subverted. The video consumption scene will be formed.
4, application point four: video editing
With the advancement of artificial intelligence, video editing has become easier, and the use of AI to significantly improve the efficiency of editing video is also of great significance.
Adobe, the world's leading provider of digital media editing software, has also joined the tide of artificial intelligence and released its first underlying technology development platform based on deep learning and machine learning, Adobe Sensei. The ability to automate the video editing process while also allowing us to control the editing style of artificial intelligence as we wish.
The system automatically organizes all the shots, including the images taken at multiple angles, in accordance with our desired script, and then finds the specified content as needed. The program can accurately identify the contents of these clips, and the system uses facial recognition and emotion recognition systems to analyze each frame. After all the elements can be organized, the system will edit and process the video according to different styles and habits, and label a certain style.
Graava, also targeting sports enthusiasts and video sharers, launched smart motion cameras and companion mobile apps. The camera has a built-in smart sensor module that recognizes the photographer's excited moments by recognizing the body's heartbeat frequency and automatically clips them into video clips. And a domestic company called Huichuan Intelligence is also a cloud platform that can quickly convert script text into short video. After the user enters an article, a link or a keyword, it will automatically search for the appropriate image and video material based on artificial intelligence technology, and with artificial intelligence synthesized speech, and finally merge into a short video.
to sum up
The above summarizes the four major application points of the computer vision and video industry, fully demonstrating that computer vision technology enables video intelligence to be continuously upgraded, Internet applications operate in video, and make video a super entry point for Internet applications. The world of pixels has extended beyond images, and while video has always been a challenge for machine learning researchers, today's technology makes it as easy to extract information from video as it is to extract information from images. The emergence of artificial intelligence, an emerging “toolâ€, has enabled humans to accelerate their journey to highly intelligent forms and reconstruct the structure and collaboration of the entire video industry.
Silicone Rubber Keypads Membrane Switches
A Silicone Rubber Keypads Membrane Switch uses compression-molded silicone rubber with conductive carbon pills or with non-conductive rubber actuators. The compression molding process creates an angled web around a keypad center. When a keypad is pressed, the webbing collapses or deforms to produce a tactile response. When the pressure on the keypad is released, the webbing returns the keypad to its original position with positive feedback. The closure of the switch circuit occurs when the conductive pill or printed conductive ink makes contact to a PCB when the web has been deformed.
Silicone Rubber Keypads Membrane Switches,4 Buttons Membrane Switch,Production Equipment Membrane Switch,Digital Control Membrane Switch
Dongguan Nanhuang Industry Co., Ltd , https://www.soushine-nanhuang.com