While millions of users upload videos to the Facebook platform daily, most of them look at the amount of interaction on those videos or their impact on people, but the company looks at it in a different way, its main goal is to find new innovative income channels as is the norm with it. It announced an artificial intelligence project called Learning from Videos that aims to automatically analyze videos to extract data that can be used to develop new products and improve the experience of ads and recommendations.
The company says it will only use publicly available videos and will not analyze any private clips that users upload to their accounts.
The mechanism of the artificial intelligence system Learning from Videos from Facebook
The new learning from videos system from Facebook divides the video clips into pictures and studies each of them to understand them and understand any existing text or movement, thus obtaining a complete analysis about the information contained in the video, and then using this information to improve artificial intelligence techniques. It owns it in order to develop new products.
The company indicates that this project will help researchers in the field of artificial intelligence to obtain data to develop their systems in a different way and without completely relying on pre-classified data, as the project will help in analyzing videos closer to what humans do and thus will contribute to obtaining more and more accurate data from methods. Ordinary.
To get the best result from the project, Facebook will analyze hundreds of videos from all countries of the world that reach the platform, due to the different interests and cultures between each region, which makes the data more accurate. Also, having accurate data from each country will help in developing more suitable products for their audience, as well as displaying more accurate advertisements.
According to Facebook, the TikTok-like Reels thumbnail feature is the first practical application for the new project, as Instagram makes recommendations and shows custom clips for each user based on their interests.
These technologies may be considered the first of their kind for American social networks, but we have previously seen them with TikTok and some Chinese companies.