The present work shows the development of a prototype system to detect the use of face masks and body temperature scanning for access control to commercial premises based on artificial vision algorithms. This system uses a quantitative field research methodology at a descriptive level by means of a model built in Tensorflow that works on the video obtained using the OpenCV2 program. The implementation consists of three modules: body temperature detection by means of the Mlx90614 sensor, detection of the correct use of the mask by means of a frontal camera analyzed with Convolutional Neural Networks and access control using electronic components. Performance tests analyze the performance metrics of the neural network with different learning rates and number of iterations such as: precision, accuracy and sensitivity, in five classification cases: high or normal temperature, faces without mask, faces with misplaced mask and with correctly placed mask. The results show a learning rate of 87.5%, displaying identifiers to detect whether a client is wearing a mask correctly and emitting sound alerts when biosecurity measures are not complied with. In relation to access control, an efficiency of 100% after detection.