基于深度学习的无人机影像车辆检测研究

Vehicle Detection Based on Deep Learning of UAV Images

本文采用分辨率较高且获取方便的无人机影像作为数据源,利用Faster R-CNN网络提取影像中车辆目标。首先用ImageNet数据集微调Faster R-CNN网络,再用无人机影像制作样本集并对Faster R-CNN网络再次训练,得到目标检测模型。将检测模型用于识别影像中的车辆目标:先提取车辆目标所在区域,排除其它地物干扰,利用分类器识别候选框中目标类别,同时使用回归器修正候选框位置。实验结果表明,Faster R-CNN网络能够有效提取影像中的车辆目标。

The UAV(unmanned aerial vehicle) image with high resolution and easy access is used as the data source, and the Faster R-CNN network is used to extract the vehicle target in the image. First, the ImageNet data set is used to tune the Faster R-CNN network, the UAV image is then used to make the sample set and the Faster R-CNN network is retrained, and the target detection model is obtained. The detection model is used to identify the vehicle targets in the image: the area of the vehicle target is extracted, the other objects are excluded, the classifier is used to identify the target category in the candidate frame,and the candidate frame position is corrected with the regressor. Experimental results show that the Faster R-CNN network can effectively extract vehicle targets in images.