Abstract: The following study presents the development of a face access control system using convolutional neural networks where all the faces of a scene are recognized and classified to allow or deny an individual access according to their facial characteristics. The system allows the access of two specific people after individually recognizing them for 5 sec in a video sequence and prevents access when the presence of an outside person, located within the area of vision is detected. The program uses a face detection system by haar classifiers, a point tracking system by KLT Algorithm (Kanade-Lucas-Tomasi) and a classification technique by convolutional neural networks where accuracy percentages are reached above 96%.