With the advent of the era of full video, all walks of life have put forward new upgrading demands for the application, experience and efficiency of video. The development of AI and cloud computing has injected sufficient impetus into the new stage of intelligent video. On May 13, Baidu Intelligent Cloud “Cloud Intelligent Technology Forum – Intelligent Video Special” activity was held in Beijing, the heavy release of intelligent video cloud 3.0 panorama. Baidu Intelligent Video Cloud 3.0 is based on the cloud native architecture and integrates Baidu’s cutting-edge video cloud technology and AI capability architecture matrix, enabling the whole process of video creation, production and application from three levels of cloud intelligence technology integration, product platform and application scenario. Hou Zhenyu baidu group vice President said in his speech, consumer Internet video content of accelerating at the same time, many sectors of the business are constantly being video, the change trend is the driving force behind the AI and the development of cloud computing, smart video experience into the content Gao Qinghua, experience a new stage of immersion, the analysis of intelligent. In order to actively respond to the trend of business video, all industries need to start to plan a unified video platform integrated with cloud intelligence. Baidu has long and rich business practice in the field of video, and based on this, it has built a leading technology capability of integrating cloud intelligence, hoping to provide a strong boost to the development of intelligent video in all industries.
Baidu Group Vice President Hou Zhenyu delivered a speech
At present, mass video collection and access, data analysis, video processing and storage have become new problems in the industry. The industry, media and consumers not only have higher expectations for video quality, efficiency and stability, but also hope to have more convenient and diversified innovative applications in video management, production and creation, interaction and form. On the forum, Baidu intelligent video cloud general manager Lian Linjiang detailed analysis of the intelligent video cloud 3.0 panorama. Intelligent video cloud 3.0 will “YunZhi one” the unique advantages of comprehensive integration and video technology, thus forming the integration of YunZhi, platform, the whole scene, provide industry with cutting-edge video cloud technology, practical and comprehensive ability of AI video matrix distribution platform of video creation and depending on the network platform, perception and intelligent video technology scheme covering the whole scene.
The basic technical architecture of cloud edge-end integration is the core support of intelligent video cloud 3.0. At present, intelligent cloud video 3.0 has built a complete cloud edge-end basic product system to further expand the computing power boundary. On this basis, intelligent video cloud 3.0 has complete technical capabilities and product matrix to meet the needs of high expansion, high stability, high performance and high efficiency. ▶ Video distribution infrastructure CDN has 1000+ nodes around the world, 100T bandwidth reserves, with strong network infrastructure to bring the ultimate video distribution experience and efficiency;
▶ Intelligent coding technology BD265 encoder extracts video features through deep learning network to generate adaptive coding parameters, and self-developed 60+ algorithms to achieve the same image quality reduce the bit rate by 30%+, increase the speed by 2x+;
▶ Intelligent Ultra HD video processing technology, through the old film repair, image quality enhancement, 4K super score and other aspects to greatly improve the image quality experience, make up for the lack of Ultra HD content, accelerate the development of the Ultra HD industry, the relevant technology has repeatedly won the title in the international authoritative competition;
▶ Real-time audio and video communication relies on multiple algorithms developed by the core to achieve smooth communication and ultra-low delay interaction experience; With the combination of full-link SRT protocol technology communication scheme and MPEGTS package, low delay live streaming can stably support multi-million level live streaming concurrency.
▶ Provide All in one smart video SDK matrix capability for different business scenarios, covering video effects, short video, interactive live streaming, player, enhanced super score, etc.
Based on the integration of cloud intelligence, the video creation and distribution platform and visual network perception platform of Baidu Intelligent Video Cloud 3.0 provide one-stop services and platform support for applications in different scenarios. The video creation and distribution platform is oriented to the pan-Internet and pan-media industries, focusing on workflow and media resource management, covering the whole process of video access, production, processing, review and analysis, and personalized recommendation and distribution. It uses the AI ability to analyze and understand data to speed up data flow and improve the utilization rate of media data. Optical networking sensing platform for traditional industries, to video terminal equipment and the unified connection and video data flow analysis and management, implement general geared to the needs of production, control and governance scenarios such as links, chain, chain pipe foundation platform, in enterprise production, industrial management, urban and social governance can play an important value. On top of cloud capabilities, AI capabilities enable smart video to move from “slash-and-burn” to “intensive” farming. Ding Errui, director of Baidu’s visual technology department, introduced the AI capabilities of “Intelligent Video Cloud 3.0 Panorama”. In terms of video content understanding and analysis, video intelligent analysis, content intelligent review and intelligent retrieval can be carried out through images, sounds, scenes, characters, text description and other information. Based on the multi-heterogeneous knowledge graph of Baidu self-research industry, which contains the largest mass knowledge of 550 billion yuan, it can enhance the deep understanding of video, improve the efficiency of video cataloging, and support video analysis, review, retrieval and other aspects to be widely used. In terms of intelligent video production, various content carriers such as graphic information, data and atlas are used to convert to short videos, and structured data and event context are enhanced based on knowledge graph to realize innovative video generation. In terms of video interaction and generation, a new AI capability matrix for interaction and generation is constructed, in which the video generation technology based on meta-learning-transfer learning can quickly customize high-quality videos of different styles according to requirements. And combining with the virtual digital image generation and speech synthesis, lip motion, such as technology, low cost can be produced rich clever, online image multiterminal covering digital people, interaction not only natural, sweet voice, can also fit a variety of business scenarios requirements, such as baidu cloud of digital virtual host “little C”, in the two sessions this year live TV presenter with excellent performance.
The forum also shared the intelligent video solutions and application cases of Baidu intelligent video cloud in the direction of pan-Internet, pan-media and pan-industry, covering scenes of interactive entertainment, content production, intelligent analysis, remote real-time communication, production management, security management and so on. In the pan-media scenario, Baidu’s intelligent video technology enables three major upgrades of the industry through the media brain platform to create an Internet ecosystem of the intellectual media industry. Intelligent video cloud combines the video technology of cloud intelligence with the media AI, media knowledge and media management, and solves the demands of intelligent video cloud technology in a variety of scenarios from the three levels of intelligent media capital management, intelligent content production and intelligent media management. In generic Internet application scenario, the content and the user is a key element to the Internet enterprise growth, baidu intelligent video technology assisted outstanding content creators, in-depth class track, with specific AI more mentally feeling super clear products to help customers reduce rate improve image quality, in the form of low-cost distribution, precision touch to reach the target audience group, forming the fission growth. Intelligent video cloud is based on Baidu’s unique advantages of cloud intelligence integrated AI capabilities, with new retail broadcast SaaS solution, online social solution, intelligent video one-stop solution, full link codec optimization solution four solutions, to meet the needs of enterprises with different requirements of customized video. In the pan-industry scenario, Baidu Intelligent Video Cloud takes the edge and end of the cloud as its infrastructure, and is dominated by platform + data to meet the needs of business docking system and data platform. The visual network perception platform in intelligent video cloud 3.0 focuses on the characteristics of industry-level video to make decisions on a variety of businesses through information acquisition. Through 200+ application scenarios covered by multiple scenes at the intelligent end, one-stop, low data, fast customization ability easyDL operator production technology, and deep integration with the cloud edge end, The optimal cloud-intelligence integrated architecture with second-level warning capability is provided to form a modular platform, and open interface services to realize rapid docking construction and decouple the allocation management of platform resources by applications. “Integration of Cloud and Intelligence” aims to better focus on the pain points and difficulties in industrial application, and play a greater role in the in-depth technological transformation and upgrading of industrial intelligence. At the event, partners from Hangzhou Xingxi Technology and Yichang Military Region shared wonderful cases of cooperation with Baidu Intelligent Video Cloud. » Star light technology co., LTD. Hangzhou cloud rhinoceros live product director Hu Yalong intelligent video Shared a joint baidu cloud for retail live digital intelligent fu to the application of the case, through baidu intelligent video cloud PaaS capabilities and cloud rhinoceros live SaaS application of combination of solving retail brands from sowing to perform threshold is high, the lack of the key issues in live stores operating force.
» yichang point military district party committee standing committee member and deputy district chief Dr Peng-fei gu said that in the share point military actively embraced and artificial intelligence, networking and the Internet of things, in baidu cloud under the participation and the power of intelligent video in public health, rural revitalization, primary health care, environmental protection for the livelihood of the people to bring more convenient and intelligent service, Let the people have a sense of security, gain a sense of happiness.
Left: Hu Yalong, director of cloud Xi live broadcasting products of Hangzhou XingXi Technology Co., Ltd
Right: Dr. Gu Pengfei, member of the standing committee and executive deputy head of Yichang Military Region
With the unique advantage of “integrating cloud and intelligence”, in addition to the direction of intelligent video, Baidu Intelligent Cloud has leading products, technologies and solutions in intelligent manufacturing, smart finance, smart city, smart energy, smart medical and other fields, and has cooperated with a number of industry leaders to promote the digitization and intelligent upgrading of enterprises. Baidu will continue to deepen its AI technology capabilities in all walks of life, lay a solid foundation for intelligence, and help achieve new breakthroughs in industrial intelligence upgrading.
