“Imagine an assistant who understands your happiness, anger, sorrow, and laughter and what you want to watch based on your feelings even before you figure out what your feelings are. It will give you the content you want most. AI can do this,” said Gong Yu, Chief Executive Officer of iQiyi, at the company’s annual entertainment marketing conference on May 18 in Beijing.
However, iQiyi says they are doing more than just the “common” application of personalized content. The internet entertainment company is developing technology to help directors cast actors and edit films. It’s building a database for actors that contains not only physical information but also their theatrical genres, behavior patterns, and clothing styles.
Algorithms are developed to identify human faces in different shows and extract their corresponding clips. For instance, if a user solely enjoys the performance of a certain actor, she can watch the clips only featuring that particular actor. Beyond that, users will also be able to select their favorite types of performances, such as singing or dancing. iQiyi has already been using this to generate featured scenes in variety shows.
The technology will appeal to business development too. “After applying facial recognition to all the videos, we can calculate the time every performer is on screen, no matter in variety shows or TV series. Longer the time is, more popular the performer is. This will be of great use to estimate his or her value in terms of promotion and marketing,” said CTO Liu Wenfeng.
The same scenario can be applied to casting. The algorithm studies an actors performances and tags it with descriptors. These are then used to match the role with the actor, saving time and money on casting. However, Liu emphasized this will only be used in selecting supporting actors and actresses because “leading roles are more dependent on the directors’ experiences and interpersonal relationships.”
iQiyi, backed by one of China’s biggest tech companies Baidu, went public on Nasdaq in late March. In its recent financial statement, it reported a 57% revenue increase compared with the same period in 2017, although it still hasn’t turned a profit. The increase is mainly drawn by membership services revenues and online advertising.
–This article originally appeared on TechNode.