The spread of COVID-19 that occurs through droplets can be avoided by reducing contact between individuals, so it is necessary to limit visitors, especially in crowded places such as shopping centers to avoid transmission between visitors. This study utilizes YOLOv3 object detection to recognize objects from camera image input, which is implemented on the Raspberry Pi 4, to identify visitors and use masks. The results of the identification of human objects will be calculated to determine the number of visitors at the shopping center. Then a buzzer sound warning is given when visitors are not wearing masks, if visitors exceed the capacity limit, they are also given a warning via an android application to the building manager. The results of the model detection show the mAP value of 77.92% for 3 classes of mask objects, without masks and humans.