Fast Vehicle Turning-Movement Counting using Localization-based Tracking
Fast Vehicle Turning-Movement Counting using Localization-based Tracking
复制标题
DOI：
10.1109/cvprw53098.2021.00469
复制
发表时间：
2021-06
期刊：
影响因子：
--
通讯作者：
Derek Gloudemans;D. Work
中科院分区：
文献类型：
--
作者：
Derek Gloudemans;D. Work
尽管交通量和转向运动数据的利用率很高，但几乎每个城市的绝大多数道路和交叉口仍然很难获得这样的数据。如果轻量级算法可以被设计成以相对适中的计算复杂性实时运行，那么边缘计算设备为记录转向运动数据提供了一个很有前途的工具。为此，本文提出了基于定位跟踪的车辆转弯运动计数(LBT-Count)算法。这种方法速度很快，因为它从不对完整的帧执行检测。取而代之的是，只有图像的一小部分被裁剪并用于检测帧内的对象。该方法在AI城市挑战赛第一赛道的公共评估服务器上获得了具有竞争力的性能(前50%的数据总体排名第七)。此外，我们还证明了LBT-Count算法比基于可用挑战数据的传统检测跟踪框架的类似计数算法快52%。
Despite the high utility of traffic volume and turning movement data, such data is still hard to come by for the vast majority of roadways and intersections in nearly every city. Edge computing devices offer a promising tool for recording turning movement data if lightweight algorithms can be designed to run in real-time with relatively modest computational complexity. To that end, this work presents Vehicle Turning-Movement Counting using Localization-based Tracking (LBT-Count). This method is fast because it never performs detection on a full frame. Instead, only a few portions of the image are cropped and used to detect objects within the frame. The method achieves competitive performance on the public evaluation server for Track 1 of the AI City Challenge (7th overall on the first 50% of data). Furthermore, we show that LBT-Count is 52% faster than an analogous counting algorithm utilizing a traditional tracking-by-detection framework on available challenge data.