A lightweight image sensitive information detection model based on yolov5s

Current sensitive information detection methods are prone to problems such as low detection accuracy, long training time, and slow detection speed, resulting in models that are usually not suitable for practical deployment. To solve this problem, a lightweight image sensitive information detection model based on yolov5s is proposed in this paper. First, this paper designs an efficient attention module GPSA module based on PSA module in the feature extraction part, which enables the network model to learn richer multi-scale feature representations and improve the detection accuracy of the model for sensitive information. In the feature fusion part, this paper adopts the BiFPN structure instead of the PAN structure of the original model, so that the feature fusion ability of the model can be improved. After experimental comparison, the results show that the detection accuracy and speed of the proposed method in this paper on the homemade sensitive image dataset are better than the current mainstream methods. The experimental results show that the final mAP of this model on the self-made sensitive image data set can reach 71%, and the detection time of a single image is 2.8ms, which can meet the requirements of network platform deployment in practical application.