Google is is testing a new feature that will help YouTubers in changing the background of a video. The new video segmentation tool will appear for YouTube Stories allowing some creators to change the background without using a green screen as is the current requirement. The feature is based on a neural network that makes it fast and easier to be used on smartphones.
The process to recognise the background of an image is difficult, which makes it more complicated for videos. According to the Google Research Blog report, a device has to do the calculation at a minimum speed of 30 frames per second for the video fragmentation of the background and foreground, which makes it difficult for smartphones, as all the smartphones are not fast enough to process data at that speed.
To solve the problem Google set up a convolutional neural network architecture and trained it on thousands of labelled images to recognise the features of a head and shoulder, a series of optimisation data needed to do so.
According to a report on Techcrunch changing the background of a video or image requires a device to recognise the foreground and background of the image or video. Recognising and processing a background can be easily done using a depth-sensing camera or device like a computer which has a faster processing speed without any constraints when it comes to battery life.
The lightweight video format is rolling out to limited beta testers and will be helpful for YouTube creators.