Integrating Weighted Feature Fusion and the Spatial Attention Module with Convolutional Neural Networks for Automatic Aircraft Detection from SAR Images

The automatic detection of aircrafts from SAR images is widely applied in both military and civil fields, but there are still considerable challenges. To address the high variety aircraft sizes complex background information images, a new fast framework based on convolution neural networks proposed, which achieves rapid with accuracy. First, airport runway areas detected to generate mask rectangular contour whole generated. Then, deep network proposed this paper, named Efficient Weighted Feature Fusion Attention Network (EWFAN), used detect aircrafts. EWFAN integrates weighted feature fusion module, spatial attention mechanism, CIF loss function. can effectively reduce interference negative samples enhance extraction, thereby significantly improving Finally, results false alarms produce final results. evaluate performance framework, large-scale Gaofen-3 1 m resolution utilized experiment. rate alarm our algorithm 95.4% 3.3%, respectively, outperforms Efficientdet YOLOv4. In addition, average test time only 15.40 s, indicating satisfying efficiency detection.