高速公路场景下基于深度学习的数据集建立

Construction of Datasets Based on Deep Learning in Highway Scenarios

本文拟建立道路交通中视频目标检测所使用的数据集,用通用的样本标注规则,完成高速公路场景中小车、卡车、大巴车和行人4类目标的标注,对不符合要求的数据进行删选,创建的数据集在不同场景下具有很好的鲁棒性。

This paper intends to establish a data set used for video target detection in road traffic,and use common sample labeling rules to complete the labeling of 4 types of targets:cars,trucks,buses,and pedestrians in highway scenes.Deletion,the created data set is very robust in different scenarios.