Abstract: Aiming to solve insufficient feature information extraction, low accuracy, and ease of deception in conventional face liveness detection, an face liveness detection based on multi-feature fusion with an attention mechanism is proposed. First, to make full use of the low-level features of the image, the original face image is transformed to obtain the Fourier Spectrum. The local binary pattern and simplified Weber local description images are used as the input of the detection model. Next, these three types of images are imported into three branches of ResNet with the same structure. Then, the attention mechanism is introduced to represent the importance of different features for recognition, and the extracted features are weighted and fused. Finally, experiments are carried out on the CASIA-FASD and Replay-Attack benchmark datasets and compared to other existing methods. The experimental results showed that the multi-feature fusion method with an attention mechanism improvs face detection accuracy and simultaneously enhancs the algorithm's robustness and generalizability.