arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1803.08460cs.CVcs.AIcs.LGcs.MM
收藏

不可见动作识别的通用表示方法

Towards Universal Representation for Unseen Action Recognition

University of California, Merced(加利福尼亚大学默塞德分校);Newcastle University(纽卡斯尔大学);Inception Institute of Artificial Intelligence (IIAI)(人工智能初始研究院)

Yi Zhu, Yang Long, Yu Guan, Shawn Newsam, Ling Shao

摘要:

未知动作识别(UAR)的目标是在没有训练样本的情况下识别新的动作类别。虽然以前的方法专注于数据集内部可见/不可见的分裂,但本文提出了一种使用大规模训练源的管道,以实现通用表示(UR),可以推广到更真实的跨数据集UAR(CD-UAR)场景。我们首先将UAR作为一个广义多实例学习(GMIL)问题,并使用分布内核从大规模ActivityNet数据集中发现“构建块”。基本的视觉和语义组件被保存在共享空间中,以实现可以有效地推广到新数据集的UR。预测的UR样本可以通过简单的语义适应来改进,然后在测试期间可以使用UR直接识别未见过的动作。在没有进一步训练的情况下,广泛的实验表明,与UCF 101和HMDB 51基准相比,有了显著的改进。

英文摘要:

Unseen Action Recognition (UAR) aims to recognise novel action categories without training examples. While previous methods focus on inner-dataset seen/unseen splits, this paper proposes a pipeline using a large-scale training source to achieve a Universal Representation (UR) that can generalise to a more realistic Cross-Dataset UAR (CD-UAR) scenario. We first address UAR as a Generalised Multiple-Instance Learning (GMIL) problem and discover 'building-blocks' from the large-scale ActivityNet dataset using distribution kernels. Essential visual and semantic components are preserved in a shared space to achieve the UR that can efficiently generalise to new datasets. Predicted UR exemplars can be improved by a simple semantic adaptation, and then an unseen action can be directly recognised using UR during the test. Without further training, extensive experiments manifest significant improvements over the UCF101 and HMDB51 benchmarks.

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。

↑