Car Security System Based on Facial Recognition With Convolutional Neural Networks

Cars as transportation objects can be classified as an expensive asset. Thus, they are also vulnerable to theft. This situation requires car owners to be vigilant and add extra security measures to prevent theft. One technological solution currently being developed extensively is biometric technology, particularly facial recognition. Human faces have unique characteristics that distinguish one individual from another. Therefore, a system that enables a car to recognize its owner's facial features could serve as an alternative security measure. This system works by storing the car owner's facial data. The car can only be started if the detected face matches the one in the database. This system employs Convolutional Neural Network and Support Vector Machine methods for facial recognition. If a valid face is detected, the computer system sends a true value to a microcontroller connected to the car's starter system. With this technology, the car's starter system can only be activated if a valid face is recognized. The research results obtained through facial image testing show an average accuracy of 83.64%. However, under certain conditions, such as well-lit images, the accuracy is higher compared to darker images. With an accuracy of 83.64%, this method is sufficiently reliable for recognizing and validating faces.