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.
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.
We introduce a novel deep-learning-based footstep localization system using ground vibrations measured using geophones and a scalable method to automatically collect training data. We show that it significantly outperforms acoustic localization.
We present a privacy-preserving localization system that uses geophones to map pedestrian locations in outdoor spaces. The system uses a 1D-CNN to compute time differences of arrival (TDOA) and achieves localization accuracy of less than 1 meter.
We develop a shipboard monitoring system using extremely low-power wearable sensors and BLE connectivity. The system measures kinematic and biometric data plus ship environmental conditions for real-time crew health tracking during emergency response scenarios.