Abstract
Aiming at the characteristics of complex background and less target pixels in UAV aerial images, a UAV aerial target detection algorithm based on improved yolov5 is proposed in this paper. First, the low-level feature map with richer small target information is introduced into the feature fusion network, so a new prediction head is added. Then, the structure of the feature fusion network is improved to realize cross layer connection and assign different weights to each feature map. Finally, the loss function is optimized to further improve the detection accuracy of the model. The model is trained with VisDrone data set. The results show that the average accuracy of the improved algorithm is improved by 4.6 percentage points compared with yolov5x, and it has better detection performance for scenes with complex background and dense small targets.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
