A Deep-Learning-Based Face and License Plate Detection Technique for The Privacy Protection in Hospi

In recent years, with the photographic lens becoming the basic equipment of the handheld device, various photos taken in public, whether it is a group photo or a landscape photo, may take pictures of nearby people and vehicles. Both the license plate and the license plate are personal privacy. Recent studies have shown that deep learning techniques can achieve excellent results in both face detection and license plate detection tasks. Therefore, this paper proposes a privacy protection system based on deep learning technology for face and license plate detection. In the face detection system, using face with large pose variations training samples in the face detection system, it is possible to predict the face of each angle through the convolutional neural network. The multi-task cascade convolutional neural network is used in the license plate detection system, and the advantages of multi-task training can be used to improve the performance of different tasks in the training phase. The face and the license plate in one image are pre-detected by the face and license plate detection system respectively, and blur the predicted face and license plate position to achieve the purpose of privacy protection. The test data in this paper is provided by GeoForce, in the hospital environments test image, the face detection accuracy rate was 96%, and the license plate detection accuracy rate was 91%.