World Models

Video Models as Simulators of Multi-Person Pedestrian Trajectories arXiv

Evaluating Video Models as Simulators of Multi-Person Pedestrian Trajectories

We propose an evaluation protocol to benchmark text-to-video (T2V) and image-to-video (I2V) models as implicit simulators of pedestrian dynamics. We use 3D reconstruction and depth estimation to extract pedestrian trajectories without known camera parameters.

October 2025 · Aaron Appelle, Jerome P. Lynch Preprint
Image-To-Video Models for Pedestrian Dynamics Simulation ICML World Models

Can Image-To-Video Models Simulate Pedestrian Dynamics?

We investigate whether image-to-video (I2V) models based on diffusion transformers can generate realistic pedestrian movement patterns in crowded public scenes by conditioning on keyframes from pedestrian benchmark datasets.