Abstract
This study presents an advanced eye-tracking system combining a Dilated Eye Network (DEN) and the Circular Hough Transform (CHT) to enhance eye movement detection. Eye-tracking, which monitors pupil dilation, eye movements, and blinking, has applications in fields such as marketing, healthcare, and human-computer interaction. Deep learning (DL) algorithms, particularly in the context of the DEN, are shown to outperform traditional methods in detecting complex visual patterns. The system was trained on eye-tracking datasets using a Deep Neural Network (DNN) and fine-tuned with pre-trained models to improve accuracy. The Circular Hough Transform was incorporated to detect circular shapes, such as eyes, from edge-detected images, improving robustness, particularly in noisy or occluded environments. This hybrid approach achieved eye detection in 17 out of 20 test images, demonstrating significant improvements in accuracy and performance compared to conventional methods. The integration of semantic segmentation and the Hough Transform further enhances detection reliability. The proposed method not only preserves image resolution and reduces computational load but also offers promising applications in eye health monitoring, especially for younger populations. Overall, this research provides a novel solution for robust, high-accuracy eye-tracking, with broad implications across multiple sectors.
