论文标题
微小的视频编解码器
Slimmable Video Codec
论文作者
论文摘要
神经视频压缩已成为一种新型的范式，结合了可训练的多层神经网络和机器学习，实现了有竞争力的率延伸（RD）的表现，但由于沉重的神经体系结构，记忆力庞大和计算需求，仍然是不切实际的。此外，通常针对单个RD权衡进行了优化模型。最近的微小图像编解码器可以动态调整其模型容量，以优雅地减少内存和计算要求，而不会损害RD性能。在本文中，我们提出了一个可靠的视频编解码器（SLIMVC），通过在微小的自动编码器中集成一个可靠的时间熵模型。尽管建筑的建筑更加复杂，但我们表明，减肥仍然是控制速率，记忆足迹，计算成本和延迟的有力机制，这都是实用视频压缩的重要要求。
Neural video compression has emerged as a novel paradigm combining trainable multilayer neural networks and machine learning, achieving competitive rate-distortion (RD) performances, but still remaining impractical due to heavy neural architectures, with large memory and computational demands. In addition, models are usually optimized for a single RD tradeoff. Recent slimmable image codecs can dynamically adjust their model capacity to gracefully reduce the memory and computation requirements, without harming RD performance. In this paper we propose a slimmable video codec (SlimVC), by integrating a slimmable temporal entropy model in a slimmable autoencoder. Despite a significantly more complex architecture, we show that slimming remains a powerful mechanism to control rate, memory footprint, computational cost and latency, all being important requirements for practical video compression.
