An improved MTCNN face detection algorithm

Abstract

Abstract: In view of the low accuracy of face detection by MTCNN, an improved MTCNN face detection algorithm is proposed. The detection accuracy is improved by the following factors: the migration learning to the original network model, the fine-tuned to the model key parameters, selected the best effect of the parameters, and determine the threshold of face confidence. Experimental results show that the improved MTCNN algorithm greatly improves the detection accuracy.