Detecting Attention and Two Different Mind Wandering States Using Blink Frequency and Emotion Analysis

Mind wandering is a cognitive state where attention shifts away from the primary task, affecting learning, productivity, and cognitive performance.We present a real-time, non-intrusive computer vision system that classifies Concentration, Deliberate (Intentional), and Spontaneous mind-wandering from eye aspect ratio (EAR) based blink dynamics and facial emotion recognition (FER), with contrast limited adaptive histogram equalization (CLAHE) for preprocessing.Our contributions are: adaptive EAR-based blink detection FER stabilization via eye angle alignment, CLAHE, and exponential moving average (EMA) smoothing; and decisionlevel temporal fusion of blink rate with FER-derived valence.Together, these advances enable robust real-time classification of concentration, deliberate, and spontaneous mind wandering, improving resistance to occlusion, noise, and illumination while maintaining temporal coherence.Experimental results demonstrate the effectiveness of this approach, with Random Forest achieving the highest classification accuracy of 99%, followed by Decision Tree at 97%.These findings indicate significant potential for applications in adaptive learning, driver monitoring, and cognitive research.