arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2006.15327cs.CVcs.LG
收藏

基于动作图的合成视频

Compositional Video Synthesis with Action Graphs

Tel Aviv University(特拉维夫大学);UC Berkeley(加州大学伯克利分校);UC San Diego(加州大学圣地亚哥分校);NVIDIA Research(英伟达研究院);Bar-Ilan University(巴伊兰大学)

Amir Bar, Roei Herzig, Xiaolong Wang, Anna Rohrbach, Gal Chechik, Trevor Darrell, Amir Globerson

摘要:

动作视频是一种复杂的信号,在空间和时间上包含着丰富的成分结构。当前的视频生成方法缺乏在多个协调的和可能同时的定时动作上调节生成的能力。为了解决这一挑战,我们建议在一个称为动作图的图形结构中表示动作,并提出新的“动作图到视频”合成任务。我们的生成模型为这项任务(AG 2 Vid)解开运动和外观功能,并通过纳入行动的调度机制,促进了及时和协调的视频生成。我们在CATER和Something-Something V2数据集上对AG 2 Vid进行了训练和评估,结果表明,与基线相比,生成的视频具有更好的视觉质量和语义一致性。最后,我们的模型展示了zero-shot的能力,通过合成新的成分的学习行动。有关代码和预训练模型,请参见项目页面https://roeiherz.github.io/AG2Video

英文摘要:

Videos of actions are complex signals containing rich compositional structure in space and time. Current video generation methods lack the ability to condition the generation on multiple coordinated and potentially simultaneous timed actions. To address this challenge, we propose to represent the actions in a graph structure called Action Graph and present the new ``Action Graph To Video'' synthesis task. Our generative model for this task (AG2Vid) disentangles motion and appearance features, and by incorporating a scheduling mechanism for actions facilitates a timely and coordinated video generation. We train and evaluate AG2Vid on the CATER and Something-Something V2 datasets, and show that the resulting videos have better visual quality and semantic consistency compared to baselines. Finally, our model demonstrates zero-shot abilities by synthesizing novel compositions of the learned actions. For code and pretrained models, see the project page https://roeiherz.github.io/AG2Video

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑