VMAF-based Bitrate Ladder Estimation for Adaptive Streaming Article Swipe
YOU?
·
· 2021
· Open Access
·
· DOI: https://doi.org/10.1109/pcs50896.2021.9477469
· OA: W3137932550
In HTTP Adaptive Streaming, video content is conventionally encoded by\nadapting its spatial resolution and quantization level to best match the\nprevailing network state and display characteristics. It is well known that the\ntraditional solution, of using a fixed bitrate ladder, does not result in the\nhighest quality of experience for the user. Hence, in this paper, we consider a\ncontent-driven approach for estimating the bitrate ladder, based on\nspatio-temporal features extracted from the uncompressed content. The method\nimplements a content-driven interpolation. It uses the extracted features to\ntrain a machine learning model to infer the curvature points of the Rate-VMAF\ncurves in order to guide a set of initial encodings. We employ the VMAF quality\nmetric as a means of perceptually conditioning the estimation. When compared to\nexhaustive encoding that produces the reference ladder, the estimated ladder is\ncomposed by 74.3% of identical Rate-VMAF points with the reference ladder. The\nproposed method offers a significant reduction of the number of encodes\nrequired, 77.4%, at a small average Bj{\\o}ntegaard Delta Rate cost, 1.12%.\n
