arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2112.04710cs.CV
收藏

Auto-X3D:通过细粒度神经架构搜索实现超高效的视频理解

Auto-X3D: Ultra-Efficient Video Understanding via Finer-Grained Neural Architecture Search

University of Texas at Austin(德克萨斯大学奥斯汀分校);Texas A&M University(德克萨斯农工大学);University of Oregon(俄勒冈大学);Picsart AI Research (PAIR)(Picsart人工智能研究院);Facebook AI(Facebook人工智能研究院)

Yifan Jiang, Xinyu Gong, Junru Wu, Humphrey Shi, Zhicheng Yan, Zhangyang Wang

摘要:

高效的视频架构是在计算资源有限的设备上部署视频识别系统的关键。不幸的是,现有的视频架构通常是计算密集型的,并且不适合于这样的应用。最近的X3 D工作提出了一个新的家庭的有效的视频模型,通过扩展一个手工制作的图像架构沿多个轴,如空间,时间,宽度和深度。虽然在概念上是在一个很大的空间中运行,但X3 D每次只搜索一个轴,总共只探索了一小部分30个架构,这并不足以探索空间。本文绕过现有的2D结构,直接在细粒度空间中搜索3D结构,其中块类型,过滤器数量,扩展比和关注块联合搜索。采用概率神经结构的搜索方法,有效地搜索在这样一个大的空间。对Kinetics和Something-Something-V2基准测试的评估证实,我们的AutoX 3D模型在类似FLOP下的准确度高达1.3%,并且在达到类似性能时将计算成本降低到x1.74。

英文摘要:

Efficient video architecture is the key to deploying video recognition systems on devices with limited computing resources. Unfortunately, existing video architectures are often computationally intensive and not suitable for such applications. The recent X3D work presents a new family of efficient video models by expanding a hand-crafted image architecture along multiple axes, such as space, time, width, and depth. Although operating in a conceptually large space, X3D searches one axis at a time, and merely explored a small set of 30 architectures in total, which does not sufficiently explore the space. This paper bypasses existing 2D architectures, and directly searched for 3D architectures in a fine-grained space, where block type, filter number, expansion ratio and attention block are jointly searched. A probabilistic neural architecture search method is adopted to efficiently search in such a large space. Evaluations on Kinetics and Something-Something-V2 benchmarks confirm our AutoX3D models outperform existing ones in accuracy up to 1.3% under similar FLOPs, and reduce the computational cost up to x1.74 when reaching similar performance.

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。

↑