The invention discloses a deep learning-based smart park parking lot parking space prediction method. The method comprises the steps of S1, collecting data of a smart park parking lot in real time; s2, analyzing image data acquired from a camera system by using a convolutional neural network, and determining an occupation state of each parking space through an image recognition technology; s3, performing time sequence analysis on the historical occupation time and frequency of the parking space by adopting a recurrent neural network; s4, comprehensively applying the processing result of the convolutional neural network and the processing result of the recurrent neural network, and constructing a hybrid deep learning model; s5, a dynamic data processing and model updating mechanism is realized, and the system adjusts a prediction strategy and model parameters according to data input in real time; s6, using a feature engineering technology to process parking space use time and position information, and integrating external factors; and S7, outputting a prediction result to a user interface of the parking management system in real time. The method has the advantages of real-time response, dynamic adaptation and high prediction accuracy.
本发明公开了一种基于深度学习的智慧园区停车场车位预测方法,包括S1、实时收集智慧园区停车场的数据;S2、使用卷积神经网络分析从摄像头系统获取的图像数据,通过图像识别技术确定每个车位的占用状态;S3、采用循环神经网络对车位的历史占用时间和频率进行时间序列分析;S4、综合应用卷积神经网络和采用循环神经网络的处理结果,构建一个混合深度学习模型;S5、实现动态数据处理和模型更新机制,系统根据实时输入的数据调整预测策略和模型参数;S6、采用特征工程技术,处理车位使用的时间和位置信息,并整合外部因素;S7、预测结果实时输出至停车管理系统的用户界面。本发明具备实时响应、动态适应和高预测准确性的优点。
Smart park parking lot parking space prediction method based on deep learning
一种基于深度学习的智慧园区停车场车位预测方法
2024-09-20
Patent
Electronic Resource
Chinese
Smart park parking space intelligent management method and system
European Patent Office | 2023
|Outdoor parking lot unoccupied parking space detection method based on deep learning
European Patent Office | 2021
|