“…Currently, target detection algorithms under neural networks can be broadly classified into two categories. The first class is the two-stage target detection algorithm represented by Faster-RCNN, which further adds suggestion and regression to one class of detection algorithms and will have better results in terms of accuracy; the other class is the single-stage target detection algorithm represented by Yolo and SSD algorithms, which skips the suggestion generation part of the traditional target detection algorithm and does not need to find the possible existence of the target in advance region, but directly gives the detection target location and target class, which has certain advantages in detection speed and efficiency [2]. In this paper, we select one of the classical algorithms among the two types of target detection algorithms, train the target detection model applicable to the dataset of this paper, respectively, and test the attacks on the detection model using the adversarial samples generated above.…”