基于改进的FAST R-CNN的前方车辆检测研究

目前,前方车辆检测的研究主要通过机器学习的方法,然而其难以解决遮挡和误检的问题。在这种背景下,使用深度学习的方法检测前方车辆更为有效。首先采用了选择性搜索方法获得样本图像的候选区域,然后使用改进的FAST R-CNN训练网络模型,检测道路前方车辆。已在KITTI车辆公共数据集上对该方法进行了测试,实验结果表明,所提方法的检测率高于CNN直接检测的结果,很大程度上解决了遮挡和误检的问题。而且,与先提取Harr-Like特征然后利用Adaptive Boosting分类器的算法相比,该方法在TSD-MAX交通场景数据库测试中实现了较高的性能。结果表明,该方法提高了车辆检测的准确性和鲁棒性。

Forward Vehicle Detection Research Based on Improved FAST R-CNN Network

The current research on vehicle detection is mainly about machine learning,but it is still difficult to deal with occlusion and false detection.In this paper,using deep learning methods to detect forward vehicles is more effective.This paper firstly adopts the selective search method to obtain the candidate area of the sample image,and then uses the improved FAST R-CNN training network to detect the forward vehicles on the road.The method has been tested in the KITTI vehicle public dataset.The experimental results show that the detection rate of this method is higher than that of the direct test based on CNN.The problem of occlusion and error detection is largely solved.Moreover,the widely used method extracts the circulated Harr-Like features,and then uses the Adaptive Boosting classifier algorithm.Compared in TSD-MAX traffic scene dataset,the proposed method provides a higher performance.The results show that this method improves the accuracy and robustness of vehicle detection.