Gaze Dataset

Gaze datasets are collections of eye-tracking data used to train and evaluate models for various applications, including gaze following, viewport prediction in 360° video, and gaze estimation for human-computer interaction. Current research focuses on improving model performance through semi-supervised learning techniques, incorporating multimodal data (e.g., video content, user history), and leveraging transformer-based architectures for better feature extraction and fusion. These advancements aim to address challenges like limited labeled data and improve the accuracy, robustness, and explainability of gaze-based systems, ultimately impacting fields such as assistive technologies, virtual reality, and medical image analysis.