Fast H.264/AVC to HEVC transcoding based on machine learning

Abstract

Since the HEVC codec has become an ITU-T and ISO/IEC standard, efficient transcoding from previous standards, such as the H.264/AVC, to HEVC is highly needed. In this paper, we build on our previous work with the goal to develop a faster transcoder from H.264/AVC to HEVC. The transcoder is built around an established two-stage transcoding. In the first stage, called the training stage, full re-encoding is performed while the H.264/AVC and the HEVC information are gathered. This information is then used to build a CU classification model that is used in the second stage (called the transcoding stage). The solution is tested with well-known video sequences and evaluated in terms of rate-distortion and complexity. The proposed method is 3.4 times faster, on average, than the trivial transcoder, and 1.65 times faster than a previous transcoding solution.