Rapid Airplane Detection in Remote Sensing Images Based on Multilayer Feature Fusion in Fully Convolutional Neural Networks

Abstract

To address the issues encountered when using traditional airplane detection methods, including the low accuracy rate, high false alarm rate, and low detection speed due to small object sizes in aerial remote sensing images, we propose a remote sensing image airplane detection method that uses multilayer feature fusion in fully convolutional neural networks. The shallow layer and deep layer features are fused at the same scale after sampling to overcome the problems of low dimensionality in the deep layer and the inadequate expression of small objects. The sizes of candidate regions are modified to fit the size of the actual airplanes in the remote sensing images. The fully connected layers are replaced with convolutional layers to reduce the network parameters and adapt to different input image sizes. The region proposal network shares convolutional layers with the detection network, which ensures high detection efficiency. The simulation results indicate that, when compared to typical airplane detection methods, the proposed method is more accurate and has a lower false alarm rate. Additionally, the detection speed is considerably faster and the method can accurately and rapidly complete airplane detection tasks in aerial remote sensing images.

Effective Airplane Detection in Remote Sensing Images Based on Multilayer Feature Fusion and Improved Nonmaximal Suppression Algorithm

“…Aiming at the problem of the insufficient representation ability of weak and small objects and overlapping detection boxes in airplane object detection, we carried out related research that has provided an original contribution to the field. We have done similar work before [2,35]. In the literature [2], we proposed an airport detection method combining cascade region proposal networks and multithreshold detection networks, which aimed to deal with the problems of complex background and inaccurate positioning.…”

“…In the literature [2], we proposed an airport detection method combining cascade region proposal networks and multithreshold detection networks, which aimed to deal with the problems of complex background and inaccurate positioning. In the literature [35], we proposed an airplane detection method to deal with the issue of small objects, which used multilayer feature fusion in fully convolutional neural networks. This research fused multilayer features by adding a maximum pooling layer, a deconvolutional operation, and a convolutional layer, but it increased the time cost of the algorithm.…”

Near-Infrared Road-Marking Detection Based on a Modified Faster Regional Convolutional Neural Network

“…The last stage (conv5) has three convolution layers, three ReLU layers, and one pooling layer [41]. Within each stage, the feature map is preserved [6,28]. After an image passes through each stage, its feature map shrinks by half.…”

“…Conversely, those features from the shallow layers are rich with precise positioning information for the object in the image but score low in semantic feature representation ability [41]. Thus, to take the advantages from the shallow layers and the deep layers and to overcome the tradeoff between the spatial resolution of the lower layers and the distinctive semantic features of the deep layers, multilayer feature fusion is imperative [28,41]. The feature maps obtained from the output of each convolution stage should have sufficient semantic and position information for an excellent performance of the model.…”

“…These feature maps are rich in semantic information but lack position information. To overcome this challenge, many researchers have presented the application of feature fusion techniques in object detection to utilize both the position and the semantic information of a feature map [27,28]. Road-marking detection is a vital aspect of the Intelligent Transportation System (ITS), and research is still ongoing in this area.…”