EgoInteract: Synthetic Egocentric Videos Generation for Interaction Understanding and Anticipation
Researchers have developed EgoInteract, a novel simulator for generating synthetic egocentric videos. This tool allows for precise control over camera movement, human actions, and object interactions within diverse environments. The generated synthetic data, complete with detailed annotations, has been used to train models that show improved performance on real-world egocentric perception tasks, demonstrating the effectiveness of simulation-based approaches for this domain. AI
IMPACT Enables more efficient training of AI models for egocentric perception tasks by providing controllable synthetic data.