论文标题
分布式基于边缘的视频分析
Distributed Edge-based Video Analytics on the Move
论文作者
论文摘要
近年来，我们目睹了数据的爆炸性增长。这些数据的大部分是由安全摄像机，智能手机和仪表板凸轮生成的视频数据。对于许多新兴应用，例如实时面部识别和对象检测，对此类数据的及时分析至关重要。在这项研究中，我们通过Dash Cam视频和现在的Edgedashanalytics（EDA）解决了实时现场视频分析的问题，这是一个基于边缘的系统，该系统可以使用本地移动设备网络实现接近实时的视频分析。特别是，它同时处理由两个或多个移动设备以几乎实时方式移动的两个或多个移动设备制作的视频。一台相机向外朝外捕获车辆前面的视线，而另一个相机向内朝内捕获驾驶员。分析外部视频以检测潜在的驾驶危险，而内部视频则用于识别驾驶员分心。 EDA通过设计和纳入多个优化，使用资源受限的瞬态移动设备实现接近实时的视频分析，并具有可耐受性损失的精度。我们已经将EDA作为一个Android应用程序实现，并使用两个仪表板凸轮和几个异构移动设备与BDD100K Dash Cam视频数据集（ARXIV：1805.04687 [CS.CV]）和DMD Driver Monitoring DataSet（Arxiv：2008.12085 [CSSSCV]）进行了评估。实验结果表明，使用有关资源约束的移动设备的转换时间和能源消耗（或电池使用），实时视频分析的可行性。
In recent years, we have witnessed an explosive growth of data. Much of this data is video data generated by security cameras, smartphones, and dash cams. The timely analysis of such data is of great practical importance for many emerging applications, such as real-time facial recognition and object detection. In this study, we address the problem of real-time in-situ video analytics with dash cam videos and present EdgeDashAnalytics (EDA), an edge-based system that enables near real-time video analytics using a local network of mobile devices. In particular, it simultaneously processes videos produced by two dash cams of different angles with one or more mobile devices on the move in a near real-time manner. One camera faces outward to capture the view in front of the vehicle, while the other camera faces inward to capture the driver. The outer videos are analysed to detect potential driving hazards, while the inner videos are used to identify driver distractedness. EDA achieves near real-time video analytics using resource-constrained, transient mobile devices by devising and incorporating several optimisations, with a tolerable loss in accuracy. We have implemented EDA as an Android app and evaluated it using two dash cams and several heterogeneous mobile devices with the BDD100K dash cam video dataset (arXiv:1805.04687 [cs.CV]) and the DMD driver monitoring dataset (arXiv:2008.12085 [cs.CV]). Experiment results demonstrate the feasibility of real-time video analytics in terms of turnaround time and energy consumption (or battery usage), using resource-constrained mobile devices on the move.
