Chinanews, Beijing, August 1 (Guo Chaokai) On August 1, the 2020 Zhijiang Cup Global Artificial Intelligence Competition was officially launched. The competition is aimed at the frontiers of scientific research and the pain points of the industry, and released four competition questions: zero-sample target segmentation, voice authentication, video generation, and unmanned vehicle safe driving simulation.
Previously, the Zhijiang Cup Global Artificial Intelligence Competition has been successfully held for two consecutive times, and has gradually received widespread attention and praise from artificial intelligence researchers and practitioners. In last year's competition, a total of more than 1,700 teams and more than 5,000 players from home and abroad signed up for the competition, most of which came from universities, scientific research institutions and well-known Internet companies in the field of artificial intelligence at home and abroad.
"Starting from the safety and credibility of artificial intelligence technology in practical applications, we designed the four major competition questions of this competition-zero-sample target segmentation, voice authentication, video generation, and driverless driving simulation. These are all Currently seeking breakthroughs in the basic algorithms of artificial intelligence, and promising technical problems in industrial innovation applications." Yuan Jixin, deputy director of Zhijiang Laboratory, said at the launch ceremony.
Chen Gang, Dean of the School of Computer Science and Technology of Zhejiang University, interpreted the four competition questions. "The Zero Sample Segmentation Challenge requires players to train models for target detection on known categories, but the models are required to be able to detect samples of unknown categories in segmented images." Chen Gang said.
"The task of this'Voice Verification Challenge' is to accurately identify audio files generated by artificially forging using AI algorithms, and at the same time to perform speaker recognition on real recorded audio files." Chen Gang said, the problem of this competition The difficulty lies in the high requirements for the algorithm's forged audio judgment and speaker recognition accuracy.
According to reports, the video generation challenge requires the algorithm model of the contestants to use intermittent single-frame images combined with descriptive text to generate single-frame images that meet the requirements, and all single-frame images that are finally generated are required to automatically synthesize a smooth and coherent video. The unmanned vehicle safety driving simulation competition is a competition question for testing unmanned vehicle safety driving algorithms and systems in response to the hot unmanned driving technology in recent years.
This year's competition has set up a number of awards, with a total of 2.6 million yuan, and there will be industry and talent policies and other related support. The competition will be open to the world, no matter individuals, universities, research institutes, enterprises or maker teams, etc., can register to participate. The algorithm competition team has a maximum of 5 people, and the technical challenge team has no limit. The preliminary round will be held in September and the final will be held in Hangzhou in October. (Finish)