The trend of Natural User Interface (NUI) has gradually accepted by global users and expanded its market demand. Eye tracking technology provided valuable NUI information on user's eye behavior and made NUI become a more effective and precise interactive interface for human research. At present, there are still many limits of eye tracking technology including costs and accuracy. Our major study projects aim to design and develop low cost but accurate eye tracker devises involving image recognition, calibration, infrared LED, natural light, and many other applications. This study used OpenSource eye tracking software with self-developed interfaces and attached a commercially available cameras outside to make eye tracker system a set of recording and analysis system. The experiment found an acceptable accuracy range of about 1.2 degrees of eye angle error rate, where high accuracy studies often not required or achieved. Low cost and better accuracy eye trackers could help NUI achieve to the next level of an innovative user interface.