【论文阅读】Video Generation from Single Semantic Label Map-CVPR2019

题目: Video Generation from Single Semantic Label Map

• 原文：https://arxiv.org/pdf/1903.04480v1.pdf
• pytorch代码：https://github.com/junting/seg2vid

摘要

任务：

video generation conditioned on a single semantic label map, which provides a good between flexibility and quality int he generation process.

从一张语义分割图，生成多帧图像，从而生成视频。

方法：

we synthesize
high quality content by only generating the first frame. We employ a cVAE for predicting optical flow as
a beneficial intermediate step to generate a video sequence
conditioned on the initial single frame.

先从语义图用image generation的方法生成图像，再用提出的方法来预测光流、再生成多帧图像。

本文用到的方法：

• image generation: pix2pixHD
• video generation: cVAE(proposed)

为什么不用video2video：