Design Of Automatic Home Door Security Using Face Recognition with A Convolutional Neural Network (CNN) Model

In Indonesia, the majority of residences continue to utilise traditional keys for door access. Keys are essential for house security; nonetheless, numerous criminals can circumvent this protection. A viable option to avert similar tragedies is to substitute the door security system with facial recognition technology. This security system employs CNN programming and an Arduino application. It utilises an ESP32 to autonomously open and close the door and engage a buzzer upon detecting a potential intruder. This research examines facial recognition technology utilised for door unlocking. The device exhibits a success rate of 71.4%, evaluated under various situations across five trials for each object. Further upgrades to the gadget and system developed from this research can be achieved by incorporating more features and tools for detecting low light levels in a room. Furthermore, the system might be engineered to detect from considerable distances, such as many kilometres away.