arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2108.08259cs.CVcs.RO
收藏

原始视频中的鲁棒人体轨迹预测

Towards Robust Human Trajectory Prediction in Raw Videos

Pennsylvania State University(宾夕法尼亚州立大学)

Rui Yu, Zihan Zhou

摘要:

由于其在自动驾驶汽车和室内机器人等应用中的重要性,人体轨迹预测最近受到越来越多的关注。然而,大多数现有的方法基于人类标记的轨迹进行预测,忽略了检测和跟踪中的误差和噪声。在本文中,我们研究了原始视频中的人体轨迹预测问题,并表明预测精度会受到各种类型的跟踪误差的严重影响。因此,我们提出了一个简单而有效的策略,以纠正跟踪失败,随着时间的推移,强制执行预测一致性。所提出的“重新跟踪”算法可以应用于任何现有的跟踪和预测管道。在公共基准数据集上的实验表明,该方法可以提高具有挑战性的现实世界场景中的跟踪和预测性能。代码和数据可在https://git.io/retracking-prediction上获得。

英文摘要:

Human trajectory prediction has received increased attention lately due to its importance in applications such as autonomous vehicles and indoor robots. However, most existing methods make predictions based on human-labeled trajectories and ignore the errors and noises in detection and tracking. In this paper, we study the problem of human trajectory forecasting in raw videos, and show that the prediction accuracy can be severely affected by various types of tracking errors. Accordingly, we propose a simple yet effective strategy to correct the tracking failures by enforcing prediction consistency over time. The proposed "re-tracking" algorithm can be applied to any existing tracking and prediction pipelines. Experiments on public benchmark datasets demonstrate that the proposed method can improve both tracking and prediction performance in challenging real-world scenarios. The code and data are available at https://git.io/retracking-prediction.

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。

↑