The invention provides a method and a system for judging illegal parking through deep learning. The method comprises the following steps of: obtaining a monitoring video stream of a preset region which at least comprises an illegal parking region; and when there is a monitored vehicle meeting the condition that the duration of staying in the illegal parking region is not less than a preset time period in the monitored video stream, judging whether a person in the monitored vehicle gets off within the preset time period or not, and if yes, judging that the monitored vehicle is an illegal parking vehicle. By using the technology, the problem of false alarm possibly existing in an illegal parking vehicle judgment rule in the prior art is solved, and the posture action estimation of the driveris added, so that the judgment rule is better met, and the judgment accuracy of an illegal parking vehicle is improved.
本发明提供一种利用深度学习判断违章停车的方法和系统,包括以下步骤:获取预设区域的监控视频流,其中预设区域至少包括违停区域;当监控视频流中存在不小于预设时间段内停留在违停区域内的监控车辆,且判断该监控车辆中的人员在该预设时间段内是否出现下车姿态,若是,则判断该监控车辆为违停车辆,技术改进了以往判定违停车辆规则中可能存在的误报问题,并增加了驾驶人的姿态动作估计,以满足并更加符合判定规则,提高了违停车辆的判定准确度。
Method and system for judging illegal parking through deep learning
一种利用深度学习判断违章停车的方法和系统
2020-12-11
Patent
Electronic Resource
Chinese
Deep learning-based illegal parking enforcement system using wide-area detection images
European Patent Office | 2022
|Road-based illegal parking management system and its illegal parking management method
European Patent Office | 2018
|