A Synthetic Eye Movement Dataset for Script Reading Detection: Real Trajectory Replay on a 3D Simulator

Researchers have developed a synthetic eye movement dataset using a 3D simulator, which can generate automatically labeled video data for training models in areas like vision-language processing and virtual reality. This approach addresses the scarcity and privacy issues associated with collecting real-world behavioral data, offering a scalable solution for developers to enhance their models' understanding of human behavior.

Read the full article at arXiv cs.CV (Vision)

---

Want to create content about this topic? Use Nemati AI tools to generate articles, social posts, and more.