Senior Technical Director
51Talk Senior Technical Director.He graduated from Zhongshan University with a master's degree from Peking University and Carnegie Mellon University.
Track: AI+Multi-Media Technology
2019/12/14 14:00 - 17:55
Image recognition, video coding, speech synthesis, network algorithm, AI is everywhere in all kinds of multi-media application scenarios, AI is the must-have weapon for all multi-media engineers.
Topics in the track
Currently more than 75% internet traffic is dominated by video, and we are expecting up to 90% of 5G traffic could be mobile video. How to compress, communicate, and analyze video contents become the key for 5G deployment and massive use. While 5G could be deployed without AI, intelligent terminal/edge/cloud will be essential to new revenues and lower costs for 5G. In this presentation, we'll talk about video compression and analysis with both traditional methods and latest AI techniques (deep learning with CNN in particular) in hardware, and how to communicate compressed video contents within 5G infrastructure.
Video often contains invalid and sensitive fragments. Traditional manual correction is time-consuming and laborious. This report will share how to corrected video by AI instead of manual labor, which achieve automatic correction in specific scenarios.
While the Audio\Video industry is booming, APPs are frequently suspended due to the spread of malicious information on nudity, violence and other naughty content. Content censorship becomes indispensable for the healthy development of Internet platforms. This talk will share some practical experience on Audio\Video content censorship with artificial intelligence techniques. It may be helpful for those who are interested in the application of audio and image algorithms to content censoring.
In resent years, machine learning, especially deep learning technologies fundamentally changed traditional speech signal processing that based on the short-time Fourier transform, and provides more noise-robust and speech-intelligible solutions. In this talk, we will discuss applying machine learning in the filed of speech enhancement, and share our experience of using it in the scenario of online education.