Subratad080808/Eye-Gaze-Estimation

This project has been implemented in three phases. At first, the face is identified in the environment and a border is made around the face. Secondly, the eye position is detected in the face. Finally, the eye gaze direction is estimated. This eye position is compared with the dataset and got a result of movement of eye direction. Predefined functions are implemented according to the necessity. The accuracy and the identification speed depend on the test dataset, training data set and the algorithm that is used for the project. Our proposed algorithm has outperformed all other state of the art results for Eye-Chimera dataset. The overall accuracies of this project are 90.21% and 99.19% for EyeChimera and HPEG datasets respectively. For EGDC, our proposed algorithm finds 86.93% accuracy.