The invention discloses a traffic signal lamp adaptive timing method based on an improved Faster-RCNN. The method comprises the following steps: carrying out manual labeling and data enhancement on collected vehicle information; a traffic information detection model based on the improved Faster-RCNN is constructed, a direction sensing module SEMA is added behind a first convolutional layer, and the learning ability for important features is enhanced; a unique convolution structure is adopted, and an attention mechanism SCB is added; the result of each layer of the FPN is transmitted into H-RPN, and a repeated bounding box is removed through EIOU-NMS; an improved region-of-interest alignment mode refROIAlign is adopted to improve the accuracy of region alignment; performing parameter compression on the Faster-RCNN model by adopting a model pruning method; an independent virtual coil is arranged for each lane, so that vehicles on different lanes can be detected. According to the method, vehicles in the intersection and small target vehicles far away from the intersection can be detected more accurately, the detection speed is high, the detection is more accurate, real-time traffic information can be provided for traffic signal lamp timing efficiently, and the traffic efficiency of the intersection is improved.
本发明公开了一种基于改进Faster‑RCNN的交通信号灯自适应配时方法,对采集到车辆信息进行人工标注以及数据增强;构建基于改进Faster‑RCNN的交通信息检测模型,在第一层卷积层后面添加方向感知模块SEMA,加强对重要特征的学习能力;并采用独特的卷积结构并添加注意力机制SCB;将FPN每层的结果分别传入H‑RPN并通过EIOU‑NMS去除重复边界框;采用改进的感兴趣区域对齐方式refROIAlign提高区域对齐的准确性;采用模型剪枝的方法对Faster‑RCNN模型进行参数压缩;对每个车道设置独自的虚拟线圈,实现对不同车道的车辆进行检测。本发明可以更精确的检测路口内的车辆以及离路口较远的小目标车辆,检测速度快并且更加精确,能够高效的为交通信号灯配时提供实时的交通信息,提高了交叉路口的通行效率。
Traffic signal lamp adaptive timing method based on improved Faster-RCNN
一种基于改进Faster-RCNN的交通信号灯自适应配时方法
2024-12-10
Patent
Electronic Resource
Chinese
IPC: | G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen |
Improved Faster RCNN for Traffic Sign Detection*
IEEE | 2020
|Agricultural pest detection algorithm based on improved faster RCNN
British Library Conference Proceedings | 2022
|Recoverable self-adaptive traffic signal lamp timing method
European Patent Office | 2022
|Ships Target Detection on Water Based on Improved Faster RCNN
British Library Conference Proceedings | 2022
|