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 2009.02406cs.CVcs.AIcs.CLcs.LG
收藏

基于自然语言的视频矩检索

Video Moment Retrieval via Natural Language Queries

Xinli Yu, Mohsen Malmir, Cynthia He, Yue Liu, Rex Wu

摘要:

在本文中,我们提出了一种新的视频矩检索(VMR)的方法,实现了最先进的(SOTA)性能的R@1度量和超过SOTA的高IoU度量(R@1,IoU=0.7)。首先,我们提出了一个多头自我注意力机制,并进一步的交叉注意力计划捕获视频/查询交互和远程查询依赖从视频上下文。基于注意力的方法可以在任意位置进行帧到查询的交互和查询到帧的交互,并且多头设置确保了对复杂依赖关系的充分理解。我们的模型具有简单的架构,可以在维护的同时实现更快的训练和推理。其次,我们还提出了使用多任务训练目标,包括时刻分割任务,开始/结束分布预测和开始/结束位置回归任务。我们已经验证了,开始/结束预测是嘈杂的,由于注释者的分歧和联合训练与时刻分割任务可以提供更丰富的信息,因为目标剪辑内的帧也被用作积极的训练样本。第三,我们建议使用早期融合的方法,它实现了更好的性能,在推理时间的成本。然而,对于我们的模型来说,推理时间不会成为问题,因为我们的模型具有简单的架构,可以进行有效的训练和推理。

英文摘要:

In this paper, we propose a novel method for video moment retrieval (VMR) that achieves state of the arts (SOTA) performance on R@1 metrics and surpassing the SOTA on the high IoU metric (R@1, IoU=0.7). First, we propose to use a multi-head self-attention mechanism, and further a cross-attention scheme to capture video/query interaction and long-range query dependencies from video context. The attention-based methods can develop frame-to-query interaction and query-to-frame interaction at arbitrary positions and the multi-head setting ensures the sufficient understanding of complicated dependencies. Our model has a simple architecture, which enables faster training and inference while maintaining . Second, We also propose to use multiple task training objective consists of moment segmentation task, start/end distribution prediction and start/end location regression task. We have verified that start/end prediction are noisy due to annotator disagreement and joint training with moment segmentation task can provide richer information since frames inside the target clip are also utilized as positive training examples. Third, we propose to use an early fusion approach, which achieves better performance at the cost of inference time. However, the inference time will not be a problem for our model since our model has a simple architecture which enables efficient training and inference.

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。

↑