Computer vision technology has been widely used in Hollywood over the past decade, due in part to the rapid advancement in standardization of technologies and applications. Although the media entertainment industry has been under pressure of change in the past year, this trend shows no signs of weakening. TV, DVD and CD-ROM are being replaced by distribution channels such as streaming media, podcasts, games and online services. This leads to fragmentation of the target group and thus to the need to meet the requirements of small target groups in a cost-effective manner. This trend is expected to intensify over the next few years through technology, development scale, investments and market competition.
The transformation of technologies and customer behavior makes it possible to gradually make media interaction smarter. Computer vision technology has effectively penetrated all connections and phases, including the intelligent production of entertainment products (content) and the personalized delivery (distribution and transmission) of end products.
<p class = "canvas-atom canvas-text Mb (1,0em) Mb (0) – sm Mt (0,8em) – sm" type = "text" content = "Content creation: Computer Vision has become and will continue to be a common technology in the industry shake in 2020 from the shackles"data-reactid =" 14 ">Content creation: Computer Vision has become and will continue to be a common technology in the industry shake in 2020 from the shackles
High-quality hardware and edge computing image processing algorithms have further improved the integration of image capture. This fulfills different pre-processing requirements (gamma and color matrix) by improving stability, reducing noise and eliminating errors. The main task of Computer Vision is to process captured images or videos in order to obtain 3D information, keyframe extraction and the associated frame assignment of corresponding scenarios. These applications have been used extensively for variety shows, sporting events and the production of streaming media content. Currently leading providers of computer vision technology are Arcsoft and Moviebook (China), Vinten, A & C, Radamec and MRMC (Great Britain), Panther (Germany) etc.
In addition to image data extraction, concept information (such as speed, distance and 3D shape) can be extracted from any image sequence. Statistics from most companies in the survey show that more and more companies are using 3D data acquisition systems to create models for animation in programs and video games. With the German X-IST, for example, animations based on facial movements can be processed. Moviebook's MCVS (Motion Capture from Video System) enables automatic tracking, distance and speed estimation, panorama reconstruction, etc.
In terms of data synthesis, it was technically feasible to synthesize background images in real time and adapt them to the viewing parameters of the camera (such as position, zoom and depth of field). This industrial demand enables more complex visual sets, including those that interact with roles and synthetic objects. This information is currently accessed in two ways: using optical or mechanical sensors attached to a remote-controlled robot camera; Use the pattern recognition in the video and then determine the position.
In terms of content cataloging, post-production workplaces will benefit from more automated and adaptable approaches, such as: B. from systems from Quantel (Great Britain), Discreet Logic Alias and Wavefront (Canada) and Moviebook (China). With these systems, 3D reconstruction, keyframe extraction, filtering, natural semantic understanding, intelligent segmentation, interpolation, etc. were introduced. The operators monitor the sequence of a sequence almost in real time and adjust the parameters to achieve a satisfactory result.
The effective management of film and video material is also an important issue for a number of broadcasters and producers. A field study has shown that image processing technology offers unprecedented benefits and is used by many companies. For example, IBM Media System indexes content based on color, texture, or shape queries. Kwai and Tik Tok implement content-based retrieval based on video frame annotations. Moviebook completes the automatic annotation of video archives. Suppliers such as Digital Vision (Sweden), Aurora and iQIYI are working on the interactive processing of low resolution content to achieve motion compensation, color correction, scratch, dirt and noise filtering, etc.
In terms of commercialization, Computer Vision has developed into a mature technology for scanning commercials and inserting advertisements in continuous frames. When broadcasting sponsored events such as football games, motor sports and variety shows, for example, a suitable video scene can be found for seamless placement of advertisements. Systems that can replace or insert such panels in real time using image processing are offered by technology leaders such as Matra Datavision (France), Orad (Israel), VDI (USA) and Moviebook (China). The resulting additional advertising revenue has developed into an important business portfolio for TV and streaming media platforms.
The proliferation of interactive and personalized technologies enabled the boom in customer-specific content in 2019. Technology leaders are looking for solutions based on AI and machine learning. They exceed the current recommendation engine (content distribution engine) and offer highly personalized products with content experience at their core. For example, interactive dramas, AI-based content scenarios and even customized roles, virtual hosts and roles. A typical example is the technical solution for virtual hosts introduced by Moviebook, with which images interact with synthetic characters, which can align them spatially and synchronize them in time. Such interactive applications are used not only in the media entertainment industry, but also in the education and VR industry.
<p class = "canvas-atom canvas-text Mb (1.0em) Mb (0) – sm Mt (0.8em) – sm" type = "text" content = "Moviebook
15811345539 "data-reactid =" 28 "> Cinema book
<p class = "canvas-atom canvas-text Mb (1,0em) Mb (0) – sm Mt (0,8em) – sm" type = "text" content = "SOURCE: Movie book "data-reactid =" 29 ">SOURCE: Cinema Guide
View source version at accesswire.com: