IoT-Enabled Camera-Based Eye Tracking and Data Analysis for Real-Time

The high cost and logistical constraints of specialized eye-tracking hardware limit its widespread application in clinical research and diagnostics. To address these limitations, a low-cost, scalable eye-tracking system is engineered to operate with a standard webcam, providing an objective tool to understand attentional processes associated with psychological conditions such as eating disorders, anxiety, and OCD. The core innovation is a dynamic calibration method that continuously maps a user's eye features to on-screen coordinates as they follow a smoothly moving target. This process generates a dense, individualized dataset of several thousand points, a significant improvement over traditional 9- or 16-point static calibrations. This rich data enables a LightGBM machine learning model to perform robust gaze prediction. The system achieves high accuracy, validated through quantitative evaluation, with a low mean absolute error (MAE) that translates to a minimal angular error. Leveraging computer vision algorithms and this novel calibration approach transforms accessible technology into a powerful, objective tool for analyzing the attentional and cognitive processes underlying various psychological conditions.