Baidu Research announced Tuesday, April 21, the launch of an AI-powered video synthesis tool Vidpress on its short video app Haokan that will allow users to produce professional video content from URLs.
Vidpress can automate the production video process from choosing clips that fit a given news topic to knitting video content with AI-synthesized narratives. After a user inputs a URL, the platform will analyze the web page using natural language understanding (NLU) models to help find matched media content, it will then enrich the story by aggregating relevant news from a wide range of sources, according to Baidu Research.
To create content, Vidpress sources images and video clips from web pages, an established media library as well as Baidu’s search engine. By using computer vision techniques like facial recognition, object detection, optical character recognition, and video understanding, Vidpress can cut and choose clips that fit the topic by analyzing the semantics of these videos, Baidu Research explained the production process on its website.
Vidpress can produce over 1,000 news videos per day, compared to 300-500 videos previously made by editors. Each clip takes up to nine to create, with an average of 2.5 minutes for a two-minute 720p video.