Sign up for 3 days! The second China artificial intelligence u00b7 multimedia information recognition technology competition invites you to fight!

category:Internet
 Sign up for 3 days! The second China artificial intelligence u00b7 multimedia information recognition technology competition invites you to fight!


Registration rules

2. Registration time: 15:00, September 17 ~ 24:00, October 26 (only three days from now, please hurry up

3. The number of teams in the same enrollment unit shall not exceed 3 (the colleges and universities and scientific research institutes will be properly released), and the number of entries of each team shall not exceed 3.

4. The materials and information provided by the teams should be accurate, true, effective, legal and complete. If any false or concealed information is found, the participants will be disqualified.

Schedule arrangement

[video depth forgery detection

The task of the competition is to identify the authenticity of the data set mixed with the deep forgery video and the real video, and compete for the detection ability of the video deep forgery. Among them, deep forgery video refers to the video generated by face reconstruction and face replacement.

[audio depth forgery detection

[variant homologous audio retrieval

The task of the competition is to retrieve the transcoding, compressed or edited audio template from the audio data to be inspected, and locate the audio template and the start time of its appearance.

[specific behavior recognition

The task of the competition is to detect and identify a certain kind of human behavior from the video, including chasing, fighting, crossing obstacles, opening packages, playing chess and card games, and group gathering behavior.

[OCR in complex scenes

Different from OCR in simple scenes with fixed patterns, such as bills, cards and official documents, the task of this competition is to detect and recognize the characters in the text images in more complex natural scenes, including both printed and handwritten characters. This paper mainly studies OCR ability in complex scenes, including two tasks: detecting text box and recognizing text content in text box.

Office of network security and Information Committee of the CPC Central Committee

Organizer

Organizer

Xiamen Bureau of industry and information technology

Xiamen Public Security Bureau

China artificial intelligence industry development alliance (AIIa)

Co Organizer

Xiamen Information Center

Institute of information engineering, Chinese Academy of Sciences

China Security Products Industry Association

Xiamen Meiya Baike Information Co., Ltd

Xiamen Bensi Information Service Co., Ltd

Beijing Baidu Netcom Science Technology Co., Ltd.

Special support unit

Peoples network

Source: editor in charge of mass news: Chen Tiqiang_ NB6485