A Street View Image Privacy Detection and Protection method based on Mask-RCNN

Abstract

The image contains many kinds of information, and the sensitive information in image is called privacy. Street view image is a type of image information widely disseminated and used on the Internet, containing a lot of private content. Private information in Street View images includes: human faces, license plates, etc. Once this type of manually collected sensitive privacy is leaked, it will cause serious privacy leakage problems. Therefore, it is necessary to study and apply how to identify and locate the privacy in street view images. The traditional privacy detection method is to manually identify the privacy target and then mosaic the private target. With the development of machine learning technology, the accuracy of automatic recognition technology has been improved, in this paper, we proposed a street view image privacy detection and protection method based on Mask-RCNN [1]. This method can automatically detect the private content in the street view image and mosaic it. Experiments show that our method has a high recognition rate for the privacy targets in street view images, and can well protect the privacy targets of street view images.