Implementation of a specified face recognition system based on video
Chao Guo, Yingyun Yang · 2019
Face recognition has been an important research direction in computer vision for a long time. The algorithms proposed in related fields are endless, and the accuracy that can be achieved is higher than ever. However, it is still quite difficult to apply face recognition technology. This paper combines the algorithms of face detection and face recognition to build a video-based face recognition system for efficient and accurate recognition of the faces of specified characters in the video. The system extracts the image of the person to be detected through MTCNN, and then uses Facenet to extract its features. Finally, the images are classified by SVM, thereby detecting the person to be detected appearing in the video. Experiments show that our method can still achieve good recognition results when the target is lacking data and the picture quality of the video is unstable. The accuracy of our method in self-built dataset in this paper can reach 94.9%.