Target detection for remote sensing based on the enhanced YOLOv4 with improved BiFPN

To solve problems for false detection, inadequate regression performance of anchor frames, and the inability to detect small targets in traditional multiscale target detection methods based on YOLOv4, we propose a novel target detection framework named as Enhanced YOLOv4. Firstly, our improved BiFPN replaced the original PANet as the feature fusion module, which can achieve multi-scale feature fusion by way of shared weights. Secondly, the channel attention mechanism (CAM) was embedded before the detection head to highlight the correlation between channels so that small targets can be get more attention. At last, to improve the anchor box regression effect and accelerate the training speed of YOLOv4, we improved the net training loss function, in which the original CIoU was replaced by CDIoU. The experimental results on the DOTA dataset validate our improvement. The mAP of our method is 90.88%, and the frame rate reached 58.76 FPS, at the same time, the speed of detection is not affected significantly.(c) 2023 National Authority of Remote Sensing & Space Science. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Automated summary

To address the issues of false detection, inadequate regression performance of anchor frames, and the inability to detect small targets in traditional multiscale target detection methods based on YOLOv4, we propose Enhanced YOLOv4. Our approach incorporates an improved BiFPN for multi-scale feature fusion, a channel attention mechanism (CAM) to highlight correlation between channels for small target detection, and a modified net training loss function with CDIoU for better anchor box regression and training speed. Experimental results on the DOTA dataset show that our method achieves a mAP of 90.88% and a frame rate of 58.76 FPS, with no significant impact on detection speed.