REMOTE SENSING TARGET DETECTION ALGORITHM BASED ON IMPROVED SSD

In the field of remote sensing image target detection, the extraction of target features and multi-scale feature fusion are the research difficulties. In order to solve the above difficulties, this paper proposes a remote sensing image object detection algorithm based on optimized SSD network. An improved RFB module was introduced into the SSD network to improve the feature extraction capability of the algorithm. A multi-scale attention feature fusion module was designed to enhance the expression ability of multi-scale feature information. The anchor box matching strategy was redesigned by using the improved K-means clustering algorithm to strengthen the network’s ability to read features. Experimental results show that the average detection accuracy mAP of the proposed algorithm reaches 95.16%, which is 16.36% higher than the original SSD algorithm, and the accuracy is significantly improved compared with other improved algorithms, which verifies the superiority of the proposed algorithm.