This article collects real-time video images of air traffic controllers’ work through a camera, uses HOG features and SVM classifiers for face detection, and then combines the ERT algorithm to accurately locate the facial feature points of the eyes, calculating the eye aspect ratio (EAR) and the percentage of time the eyes are closed within a unit of time (PERCLOS). By correlating with the subjective fatigue level of the Stanford Sleepiness Scale (SSS), it is found that there is a significant positive correlation between PERCLOS and the level of fatigue, with a correlation coefficient as high as 0.859. In addition, through ROC curve analysis, the AUC value of PERCLOS reaches 0.996, with a 95% confidence interval of 0.906~1.000, showing that this detection system has extremely high accuracy and stability in distinguishing between awake and fatigue states. This study provides an effective tool for detecting fatigue in the field of air traffic control, which helps to improve flight safety and work efficiency.