Small target detection based on improved YOLOv7 remote sensing image
At present,the target detection technology has become mature,but small target detection is still a difficult point in research,especially for remote sensing images with small targets and complex backgrounds.In order to solve this problem,an im-proved YOLOv7 target detection model was proposed.Combined with the idea of multi-scale feature fusion,a weighted feature pyra-mid network BiFPN is added in the feature extraction stage of YOLOv7,and BiFPN fuses feature maps of different scales to obtain richer feature representations,thereby improving the accuracy of object detection.In addition,the BiFormer attention module is used to improve the sensitivity of the network to small-scale targets,reduce the influence of noise,optimize the detected targets,and improve the detection efficiency.Finally,the MPDIoU loss function is used to solve the limitations of the CIoU loss function to im-prove the generalization of the model.Experimental results show that the mAP value of the improved model is 4.1 percentage point higher than that of the original YOLOv7 model on the VisDrone dataset,and the detection accuracy is also improved.
万方数据
