Hierarchical Autoregressive Modeling for Neural Video Compression
Hierarchical Autoregressive Modeling for Neural Video Compression
复制标题
DOI：
--
复制
发表时间：
2020-10
期刊：
影响因子：
--
通讯作者：
Ruihan Yang;Yibo Yang;Joseph Marino;S. Mandt
中科院分区：
文献类型：
--
作者：
Ruihan Yang;Yibo Yang;Joseph Marino;S. Mandt
Marino等人（2020）最近的工作表明，通过将掩蔽自回归流与分层潜变量模型相结合，序列密度估计的性能得到了改善。我们画了这样的自回归生成模型和有损视频压缩的任务之间的连接。具体来说，我们查看最近的神经视频压缩方法（Lu等人，2019年; Yang等人，2020 b; Agustssonet al.，2020）作为广义随机时间自回归变换的实例，并提出了基于这种见解的增强途径。对大规模视频数据的综合评估表明，与最先进的神经和传统视频压缩方法相比，该方法具有更好的率失真性能。
Recent work by Marino et al. (2020) showed improved performance in sequential density estimation by combining masked autoregressive flows with hierarchical latent variable models. We draw a connection between such autoregressive generative models and the task of lossy video compression. Specifically, we view recent neural video compression methods (Lu et al., 2019; Yang et al., 2020b; Agustssonet al., 2020) as instances of a generalized stochastic temporal autoregressive trans-form, and propose avenues for enhancement based on this insight. Comprehensive evaluations on large-scale video data show improved rate-distortion performance over both state-of-the-art neural and conventional video compression methods.