Real-time face recognition method based on MTCNN-Inception-ResNet-v2-SVM model

With the widespread use of smart devices such as smartphones, facial recognition applications have experienced rapid growth. Additionally, breakthroughs in deep learning algorithms have led to the development of facial recognition technology based on deep convolutional networks, greatly improving recognition speed and accuracy. Therefore, this study proposes a real-time face recognition method based on the MTCNN-Inception-ResNet-v2-SVM model. In the facial detection phase, the MTCNN algorithm, which offers better overall performance compared to face detection algorithms such as OpenCV and Dlib, is utilized. Data augmentation and other methods are employed in the image preprocessing phase to enhance data diversity. The Inception-ResNet-v2 deep convolutional neural network is used as the backbone network for feature extraction in the facial recognition backbone network section. In the final classification stage, an SVM classifier is employed for the ultimate facial classification. Comparative analyses with models such as Inception-ResNet-v1, Inception-v3, and Inception-v4 are conducted in the backbone network section, determining that the Inception-ResNet-v2 model exhibits the best overall performance. The final result is a real-time face recognition model, MTCNN-Inception-ResNet-v2-SVM, with a high accuracy of 98.79% and fast processing speed.