Technological developments are increasing very rapidly nowadays, there are various up-to-date and up-to-date technologies in various facilities, such as IoT-based security cameras. A facial image taken using a camera, there are some disturbances that may occur caused by the imperfect shooting process so that it will be difficult to recognize. Images that are difficult to recognize require Image Denoising to make images more easily recognized. Image Denoising aims to suppress noise in the image. Noise is an error that occurs in the image retrieval process which causes a pixel intensity value to not reflect the actual pixel value. This study aims to determine whether there is an increase in the level of face detection in noise-affected images. Depending on the speed and lightness of computation, this is an important consideration. This study utilizes You Only Look Once (YOLO) for face detection because the YOLO method does not require too many datasets to recognize faces properly when compared to using the Single Shot Detector (SSD) method which requires more dataset references. many. The results of this study are models that have been successfully trained to produce input images that have noise, which have increased accuracy. Evidenced by the Precision value without Gaussian Filter of 73.39% to 85.62% after adding Gaussian Filter. The Recall value without Gaussian Filter is 80.86% to 90.55% after adding Gaussian Filter. The previous accuracy without the Gaussian filter was 69.53%, which increased to 78.6% after adding the Gaussian filter.