2 papers across 2 sessions
We introduce TrajICL, a novel in-context learning (ICL) framework for pedestrian trajectory forecasting that enables rapid adaptation without costly fine-tuning.