Development of a solution to assess the quality and to optimize AI-based video codecs

Current video codecs consider algorithms to analyze video imagery in order to find out which bits can be removed for file size reduction without subjective video frame degradation. Integrating AI with encoding process improves the quality of encoding and decoding. AI permits the software to proactively assess the quality of the encoded video before transmission. This allows the compressing system to detect and remedy any possible artifacts in the video frames. The main objectives of the company regarding this project can be summarized as 1) Quality assessment regarding the AI-based video codecs from the industry perspective. 2) Determination of codec capability of running on off-the-shelf hardware. 3) Evaluation of AI-based codecs optimization techniques such as vectorization, SIMD, GPU, etc. The enhancement in such technologies could improve the entertainment industry in the country and also potentially assist smart city projects in Canada.

Vahid Khorasani-Ghassab
Faculty Supervisor: 
Nizar Bouguila
Partner University: