Object Detection of Remote Sensing Airport Image Based on Improved Faster R-CNN

Abstract

In order to effectively improve the detection accuracy of remote sensing images in airport areas, basing on the representative deep network Faster R-CNN as the object detection method, a deeper basic network ResNet and feature fusion component FPN are used to extract more robust deep distinguishing features, and add a new fully connected layer to the end detection network and combine the softmax classifier and 4 logistic regression classifiers for object detection according to the inter-class correlation of the object. Experiments show that the improvement of the original network brings a 7.7% mAP improvement to 76.6% of the mAP. Compared with other mainstream networks, it also has a better accuracy rate. At the same time, by appropriately reducing the input amount of the proposals, the speed can be increased 3 times to 0.169s under the premise of reducing the accuracy by 2.2%. According to the specific task, the accuracy and detection speed can be reasonably weighed, which reflects the effectiveness and practicability of the network.