With the development of information society, the biometric technology characterized by iris and face plays an important role in ensuring information security. However, the traditional iris recognition collection distance is small, the cost is high, so that its civilian promotion is restricted. And the multi-modal method of combining iris and face is affected by the new crown epidemic, and the face is blocked by the mask, resulting in a decrease in recognition rate. Therefore, this project has developed a dual-modal fusion identification system based on deep learning, which provides the possibility to achieve high-precision identification while reducing equipment requirements. This method uses the characteristics of iris with multi-feature reading point and eye circumference with large size to relax the camera restriction, and combines the traditional iris feature recognition with the new biometric eye circumference recognition, which provides a robust and complementary new method for multi-modal biometric recognition technology in the epidemic environment.