Time-stamp aware anomaly detection in traffic videos is an essential task for the advancement of the intelligent transportation system. Anomaly detection in videos is a challenging problem due to sparse occurrence of anomalous events, inconsistent behavior of a different type of anomalies and imbalanced available data for normal and abnormal scenarios. In this paper, we present a three-stage pipeline to learn the motion patterns in videos to detect a visual anomaly. First, the background is estimated from recent history frames to identify the motionless objects. This background image is used to localize the normal/abnormal behavior within the frame. Further, we detect an object of interest in the estimated background and categorize it into anomaly based on a time-stamp aware anomaly detection algorithm. We also discuss the challenges faced in improving performance over the unseen test data for traffic anomaly detection. Experiments are conducted over Track 3 of NVIDIA AI city challenge 2019. The results show the effectiveness of the proposed method in detecting time-stamp aware anomalies in traffic/road videos.
Good Practices and A Strong Baseline for Traffic Anomaly Detection
Baidu Inc（百度公司）
交通异常检测是智能城市交通管理系统的重要组成部分。以往的研究工作提出了许多值得注意的见解，并在这一领域向前迈出了一步，然而，处理复杂的交通环境仍然是一个挑战。此外，缺乏高质量的数据和交通场景的复杂性，促使我们从手工制作的角度来研究这个问题。在本文中，我们提出了一个简单而有效的框架，包括预处理，动态跟踪模块，和后处理。通过视频稳定、背景建模和车辆检测，预处理阶段旨在生成候选异常。动态跟踪模块利用车辆的运动模式和时空状态来寻找和定位异常的开始时间。最后，我们使用后处理来微调异常的时间边界。毫不奇怪，我们提出的框架在NVIDIA AI CITY 2021流量异常检测排行榜中排名$1^{st}$。该代码可从以下网址获取：https://github.com/Endeavour10020/AICity2021-Anomaly-Detection。
MEDAVET: Traffic Vehicle Anomaly Detection Mechanism based on spatial and temporal structures in vehicle traffic
Universidade de São Paulo（圣保罗大学）；State University of Southwest Bahia（巴伊亚西南州立大学）；University of Manchester（曼彻斯特大学）；Brock University（布罗克大学）；Institute of Mathematical and Computer Sciences, University of São Paulo（圣保罗大学数学与计算机科学研究所）
本文提出 MEDAVET 机制，通过计算机视觉检测、二分图与凸包跟踪车辆，并结合 QuadTree 和遮挡车辆结构检测高速公路交通异常，在 Track4 上取得 F1 85.7%。
Microscopic Vehicle Trajectory Datasets from UAV-collected Video for Heterogeneous, Area-Based Urban Traffic
Transportation Research & Injury Prevention Centre (TRIPC), Indian Institute of Technology Delhi（印度理工学院德里分校交通研究与伤害预防中心（TRIPC））；Indian Institute of Technology Delhi（印度理工学院德里分校）；Queensland University of Technology（昆士兰科技大学）
