Gaze Estimation Using VGG16 Architecture Through XGBoost Classifier
Dhanvanth Reddy Yerramreddy, Veerababu Addanki, Sathvik Durgapu, Stredney Don · 2023
Gaze Estimation, a technology that has evolved in the modern era that revolves around analysing the movement of eyes and their patterns which covers a wide array of fields and domains. It transformed human-computer interaction by providing hands-free device control and navigation through visual focus. In this study, an unique approach has been introduced for estimating the gazes of the eyes through VGG16 model by modifying it’s architecture that preserves it’s base layers but replacing the softmax layer with an XGBoost classifier for better classification. For estimating it’s efficiency, a comparative analysis has been conducted with various deep learning architectures. Through the implementation of the architecture, the primary aim is to outperform the benchmark models with notable performance improvements. The outcomes are highly encouraging and the modified architecture provides more precise estimation of gazes than the benchmark models.