基于Faster R-CNN的高速公路拥堵场景车辆目标检测研究

Vehicle Object Detection Method in Highway Congestion Based on Faster R-CNN

为解决通用目标检测算法对于密集车辆检测的误检和漏检等问题,本文提出了一种基于Faster R-CNN的高速公路拥堵场景车辆目标检测方法。实验提出多变化处理模块和排斥力损失,在补充数据多样性的同时提高模型的泛化能力,并针对密集场景车辆遮挡等情况,提高密集车辆的检测精度。实验结果表明,该网络模型相比现有检测方法达到了更好的检测效果。

To solve the problem of false detection and missed detection of dense vehicle detection by general object detection algorithms,a vehicle object detection method based on Faster R-CNN in highway congestion was proposed.The proposed multichange processing module and a repulsion loss,which improves the generalization ability of the model while supplementing the di⁃versity of data,and improves the detection accuracy of dense vehicles for situations such as vehicle occlusion in dense scenes.The experimental results show that the network model achieves better detection results than existing detection methods.