Baidu Intelligent Cloud and the "Mystery of Infinity" program reached the growth line of technical cooperation

The U.S. Stock Research Agency learned that according to the news of the Titanium Media App on February 8, Baidu Group recently announced that Baidu Intelligent Cloud and CCTV's "Mystery Infinite" program reached a technical cooperation line to bring scientific and technological assistance to the program presentation of competitive sports and the daily training of athletes with "3D + AI" technology.

It is reported that through the "3D + AI" technology based on Baidu Intelligent Cloud, Baidu Intelligent Cloud has created a 1:1 3D model for the Shougang Ski Jump, and the audience can achieve "science and technology viewing" through technologies such as "same field competition", "time and space freeze-frame", "3D scene roaming" and so on. In addition, the technology has been applied in diving, table tennis training, to create an "AI coach" for the national team.

In the "Infinite Mysteries" program broadcast by CCTV-1 Comprehensive Channel and CCTV-16 Olympic Channel, Wu Tian, vice president of Baidu Group, revealed to the audience how to present competitive sports to the audience more graphically and popularly through "3D +AI" technology. Baidu Intelligent Cloud freezes the continuous action of the time series into 3D space through the "space-time freeze-frame" technology, realizing 360-degree multi-angle viewing.

Wu Tian said that Baidu Intelligent Cloud established a "same-field competition" system through "3D + AI" technology, turning the single-player competition project into a "multi-person competition", realizing the three-dimensional recovery and virtual superposition of the first and second runner-up competition screens, which is convenient for the audience to see the real-time actions of different players through a track. At the same time, through technical means, the athletes' movements are quantitatively analyzed, and a series of motion data such as gliding speed, lifting height, landing distance, and rotation angle are superimposed on the original picture. This solves the problem of high-speed, high-difficulty action restoration.