The level of security in terms of access is one of the main priorities of everyone to improve the security system that feels the need for improvement following the development of modern technology. This study discusses a security system using Android-based face recognition. The aim of this research is that the safe safety system has a better level of security than the previous system. The initial stage to build this system, the authors do the literature data collection stage as a basis for the theory and system development methods used by software designers before is the waterfall method, in general this method is divided into several stages, including: Analysis, Design, Program Code and Unit Testing. For the method used in the research of this system is the eigenfaces algorithm method for the detection of facial objects in the initial process of image training. As well as the Local Binary Patterns algorithm method and Histrogram Equalization at the stage of reading the user's face recognition image accurately which has an accuracy of face reading up to 95.56%. The results of the user's face data will be processed in Wemos D1 and the data will be sent and stored in a database. The results of data from face recognition data will be used again as user data to open the safe. The conclusion, the system can read the user's face in real time and can work well for safe security systems