arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.15796cs.CV
收藏

AR-Net:用于有效动作识别的自适应帧分辨率

AR-Net: Adaptive Frame Resolution for Efficient Action Recognition

MIT-IBM Watson AI Lab(麻省理工学院-IBM沃森人工智能实验室);IBM Research(IBM研究院);Massachusetts Institute of Technology(麻省理工学院);Boston University(波士顿大学)

Yue Meng, Chung-Ching Lin, Rameswar Panda, Prasanna Sattigeri, Leonid Karlinsky, Aude Oliva, Kate Saenko, Rogerio Feris

摘要:

动作识别是计算机视觉中一个开放的、具有挑战性的问题。虽然目前最先进的模型提供了出色的识别结果,但其计算费用限制了其对许多现实应用的影响。在本文中,我们提出了一种新的方法,称为AR-Net(自适应分辨率网络),该方法根据输入为每个帧选择最佳分辨率,以便在长的未修剪视频中进行有效的动作识别。具体而言,给定视频帧,使用策略网络来决定动作识别模型应该使用什么样的输入分辨率进行处理,目标是提高准确性和效率。我们有效地训练的政策网络与识别模型,使用标准的反向传播。在几个具有挑战性的动作识别基准数据集上进行的大量实验很好地证明了我们所提出的方法优于最先进方法的有效性。该项目的页面可以在https://mengyuest.github.io/AR-Net上找到

英文摘要:

Action recognition is an open and challenging problem in computer vision. While current state-of-the-art models offer excellent recognition results, their computational expense limits their impact for many real-world applications. In this paper, we propose a novel approach, called AR-Net (Adaptive Resolution Network), that selects on-the-fly the optimal resolution for each frame conditioned on the input for efficient action recognition in long untrimmed videos. Specifically, given a video frame, a policy network is used to decide what input resolution should be used for processing by the action recognition model, with the goal of improving both accuracy and efficiency. We efficiently train the policy network jointly with the recognition model using standard back-propagation. Extensive experiments on several challenging action recognition benchmark datasets well demonstrate the efficacy of our proposed approach over state-of-the-art methods. The project page can be found at https://mengyuest.github.io/AR-Net

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。

↑