arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2203.16062cs.CVcs.IR
收藏

AxIoU:一种公理化的视频矩检索方法

AxIoU: An Axiomatically Justified Measure for Video Moment Retrieval

Cyberagent, Inc.(赛博艾坚特公司);Waseda University(早稻田大学);Osaka University(大阪大学);Tampere University(坦佩雷大学);University of Oulu(奥卢大学)

Riku Togashi, Mayu Otani, Yuta Nakashima, Esa Rahtu, Janne Heikkila, Tetsuya Sakai

摘要:

评价措施对研究方向有着至关重要的影响。因此,对于常规方法不太适用的新应用,开发适当和可靠的评估方法至关重要。视频矩检索(VMR)就是这样一种应用,并且当前的实践是使用R@ $K,θ$来评估VMR系统。然而,这一措施有两个缺点。首先,它是排名不敏感的:它通过将列表视为集合而忽略排名前$K$的列表中成功定位的时刻的排名位置。其次,它使用阈值$θ$将每个检索到的视频时刻的交集(IoU)二值化,从而忽略排名时刻的细粒度定位质量。我们提出了一种替代措施,用于评估VMR,称为平均最大IoU(AxIoU),这是免费的上述两个问题。我们表明,AxIoU满足两个重要的公理VMR评估,即\textbf{Invariance against Redundant Moments}和\textbf{Monotonicity with respect to the Best Moment},也R@ $K,θ$满足第一个公理。我们还从经验上研究了AxIoU与R@ $K,θ$的一致性,以及它在测试数据和人类注释的时间边界变化方面的稳定性。

英文摘要:

Evaluation measures have a crucial impact on the direction of research. Therefore, it is of utmost importance to develop appropriate and reliable evaluation measures for new applications where conventional measures are not well suited. Video Moment Retrieval (VMR) is one such application, and the current practice is to use R@$K,θ$ for evaluating VMR systems. However, this measure has two disadvantages. First, it is rank-insensitive: It ignores the rank positions of successfully localised moments in the top-$K$ ranked list by treating the list as a set. Second, it binarizes the Intersection over Union (IoU) of each retrieved video moment using the threshold $θ$ and thereby ignoring fine-grained localisation quality of ranked moments. We propose an alternative measure for evaluating VMR, called Average Max IoU (AxIoU), which is free from the above two problems. We show that AxIoU satisfies two important axioms for VMR evaluation, namely, \textbf{Invariance against Redundant Moments} and \textbf{Monotonicity with respect to the Best Moment}, and also that R@$K,θ$ satisfies the first axiom only. We also empirically examine how AxIoU agrees with R@$K,θ$, as well as its stability with respect to change in the test data and human-annotated temporal boundaries.

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。

↑