Image Processing Based Lie Detector Model

Image processing refers to information processing techniques used with computers integrated into images in various industrial and technological fields. These techniques can be used in various applications involving the process of object recognition and are developed within the scope of artificial intelligence. In recent years, image processing techniques have been utilized in facial recognition systems to detect crime and criminals. This paper presents a study on the creation of a lie detector that provides information on whether individuals are lying by using image processing techniques. For this purpose, algorithms that analyze eye blinking and emotional state behaviors from the facial region detected in the video image for lie detection were developed. These algorithms were coded in the Python programming language and combined to obtain a lie detector model. Additionally, a balanced test set consisting of video images of individuals telling the truth and lying were prepared to validate the model. In tests conducted on this set, the success rate of the obtained lie detector model was found to be 83.3%.