The Google research team recently published the article 'Mobile Real-time Video Segmentation' on the Google Research Blog, stating that it has developed a new feature of video segmentation that can directly capture the foreground of the movie when shooting a movie. Etc.) Separated from the background and replaced with other backgrounds. According to Next Reality reports, many film directors or video creators will use video segmentation to divide the foreground and background of the video into two layers and enhance it by replacing the background. The effect of the film, but traditionally it takes time-consuming and laborious post-production, or requires the use of a green screen as a background to capture movement for post-production. The Google research team uses the mainstream technology of machine learning convolutional neural networks ( Convolutional Neural Networks (ConvNets) simplifies this time-consuming task by allowing the film to be split directly between the foreground and the background during instant filming. Google also invited YouTubers to use this new feature to make a video and upload it to YouTube for testing. YouTube stories for users to watch. Google leader and software engineer Valentin Bazarevsky and Andr of the project In the article, ei Tkachenka stated that the current goal is to observe the effects of some of the videos created by the creators using this new feature, and will continue to add more 'labels' to the scene to allow users to use more different ones in the future. The background, and plans to integrate into the larger Google Augmented Reality (AR) service. In fact, Apple had previously seen the world of instant video segmentation technology through Apple Clips, but this requires the iPhone X and its TrueDepth camera. Execution, Google's computer vision also allows the public to obtain a similar experience, in fact the team designed ConvNets is to solve the tedious film splitting technology and easily show high-definition video on the phone, this method can run on the iPhone 7 Videos that exceed 100 frames per second (FPS) are over 40 FPS on Pixel 2.
