The invention relates to the technical field of video processing, artificial intelligence and deep learning, and particularly discloses a vehicle illegal parking real-time detection method in a community monitoring scene, which comprises the following steps: collecting a vehicle data set in a community environment, and training a vehicle detection network model VDM-CNN designed by the invention until parameters are optimal; using the trained VDM-CNN network model to carry out vehicle target detection on a no-parking area in the community, and using a tracker to store detected target vehicle information; calculating a vehicle information relationship in the tracker, judging whether the same vehicle performs time and space relationship calculation on the same vehicle in a time sequence or not, judging whether a target has time continuity and space invariance or not, and further judging whether illegal parking behaviors exist or not; and enabling a GPU scheduling strategy to perform GPU scheduling. The invention has high detection accuracy, analyzes and judges whether the vehicle in the no-parking area is illegally parked or not from two perspectives of time continuity and space invariance, and has good effectiveness and robustness.
本发明涉及视频处理、人工智能、深度学习技术领域,具体公开了一种社区监控场景下车辆违停实时检测方法,包括以下步骤:社区环境下采集车辆数据集,训练本发明设计的车辆检测网络模型VDM‑CNN,直至参数最优;使用训练好的VDM‑CNN网络模型,对社区内禁停区域进行车辆目标检测、使用跟踪器存储检测到了目标车辆信息;计算跟踪器内车辆信息关系,判断是否有相同车辆对相同车辆在时间序列上进行时间、空间关系计算,判断目标是否具有时间连续性和空间不变性,进而判断是否有违规停车行为;GPU调度策略进行GPU调度。本发明具有较高的检测准确率,从时间连续性和空间不变性两个角度对禁停区域内的车辆是否违规停车进行分析判断,具有很好的有效性和鲁棒性。
Vehicle illegal parking real-time detection method in community monitoring scene
一种社区监控场景下车辆违停实时检测方法
2021-01-29
Patent
Electronic Resource
Chinese
Electric vehicle illegal parking real-time detection platform
European Patent Office | 2021
Vehicle illegal parking identification method for complex application scene
European Patent Office | 2021
|Vehicle illegal parking detection method based on urban high-complexity detection scene
European Patent Office | 2023
|