Investigation of Aircraft Target Detection of Remote Sensing Images Based on the Improved YOLOv5

An improved YOLOv5 algorithm for aircraft target detection in remote sensing images is proposed by this paper, of which addresses the issues of misdetection or false detection caused by dense arrangement of aircraft targets, complex background conditions, diverse attitude angles, and numerous small targets. Firstly, the coordinate attention mechanism module is added to the backbone network of feature extraction to make it easier to find aircraft targets and reduce the influence of clutter information. Secondly, by improving the multi-scale feature fusion layer, the algorithm improves the feature extraction ability and global perception ability of aircraft target. Finally, EIoU is used as a loss location function to improve the regression accuracy. By using DOTA and DIOR's joint aircraft dataset, AP, AP50, AP75, APs, AP
M
and AP
L
have respectively increased 3.6%, 3.1%, 4.9%, 3.6%, 4.5% and 1.0% compared with the original YOLOv5 algorithm, and the detection FPS has reached 34.5. The experimental results show that this proposed algorithm is suitable for aircraft target detection in remote sensing images, and can effectively improve the accuracy of aircraft detection and reduce the false detection rate.