Deteksi Mata di Video Smartphone Menggunakan Mediapipe Python

Abstract

Eye detection technology is used to recognize and analyze unique features of a person's eyes as a way to identify or authenticate their identity. This technology can be used in various applications such as pattern recognition, biometric systems, surveillance systems, and others. Most applications require precision in eye detection, so a fast and reliable eye detection method is needed. In this research, an eye detection method is proposed using the Python OpenCV and MediaPipe libraries, which offer better accuracy compared to existing solutions. Both libraries are implemented in the Python programming language, which is popular among software developers for its ability in object-oriented programming, easy data manipulation and processing, and availability of libraries and modules in various fields such as artificial intelligence. The system was tested using videos captured using a smartphone. Although the videos were captured under suboptimal conditions, such as imperfect lighting, testing was conducted on 56 videos that had relatively good quality and lasted about 5-10 seconds. The results obtained showed an accuracy rate of 100%. Additionally, the system can distinguish between open and closed eye conditions, which will facilitate further research in detecting eye blinks. In conclusion, the model created can detect eyes with a very high accuracy rate.