关键词: 智能交通, 路侧感知, 数据集, 多源感知, 激光雷达, 目标识别

Abstract:

To meet the need of standard open datasets for the researches of roadside multi-source fusion sensing algorithms， this paper proposed a method to construct a standard roadside multi-source sensing dataset. The LIDAR and image data was collected at an urban T-junction and matched each other in both spatial and temporal dimensions. A vehicle three-dimension configuration extraction method was proposed， which included the steps of road space division， road pavement segmentation， and laser point cloud clustering. A vehicle labeling method was designed， which included the steps of target filtering and classification， recognition difficulty division， 3D bounding box calibration， and tag information supplement. It constructed a standardized roadside multi-source sensing dataset that contained the labels of 9 794 cars and heavy vehicles in daylight and nighttime. The YOLOv5 algorithm and the PointRCNN algorithm were used to test the 2D and 3D target recognition performance on the constructed dataset. Test results showed that due to the differences of scene complexity， data collection device and vehicle type， the constructed dataset and open vehicular datasets had differences in terms of average number of scene and vehicle laser points， and size of vehicle 3D bounding box. The YOLOv5 algorithm and the PointRCNN algorithm have similar vehicle target recognition accuracy on the open vehicular datasets and the constructed roadside multi-source sensing dataset.

Key words: intelligent transportation, roadside sensing, dataset, multi-source sensing, LIDAR, target recognition