ReWorld-Track: A Recursive Event World Model for Language-Guided Multi-Camera Tracking

2026-09-29

AI Summary

This study addresses the issues of identity error propagation and uncertainty loss caused by early erroneous associations in cross-camera tracking. To this end, it proposes a Recurrent Event World Model that leverages recurrent neural networks to construct persistent belief states, preserving matching candidate uncertainties while dynamically updating beliefs via Bayesian posterior estimation. Furthermore, by integrating language-guided feature fusion with multimodal evidence association, the method explicitly incorporates association uncertainty into future predictions, effectively maintaining trajectory diversity and guiding identity decisions. Experimental results demonstrate that the proposed approach achieves HOTA scores of 65.19 and 45.36 on the CityFlowV2 and MTMMC datasets, respectively, significantly improving both identity continuity and prediction accuracy.

Abstract

Language-guided multi-camera tracking must preserve a target identity across unobserved gaps, where similar candidates and uncertain returns can make early associations unreliable. A wrong match can corrupt the history used to predict later observations and propagate identity errors across subsequent camera handoffs. We propose ReWorld-Track, a recursive event world model that carries association uncertainty into future predictions. Candidate matches and continued waiting define alternative target states, whose posterior probabilities are used to update a persistent recurrent belief. This representation preserves uncertainty about alternative trajectories through successive observations. This belief predicts the next camera, arrival time, and entry region, while appearance and language evidence guide association. By training across successive handoffs, the model learns to retain uncertainty that remains useful for later predictions and identity decisions. ReWorld-Track achieves HOTA scores of 65.19 on CityFlowV2 and 45.36 on MTMMC, with improved identity continuity across repeated handoffs. On MTMMC, its structured posterior update gains 0.50 HOTA points over a similarly sized generic updater and 0.94 points over fixed-moment soft association, raising next-camera accuracy from 86.03% to 87.41% and reducing median arrival-time error from 0.78 s to 0.71 s for subsequent target returns.

Problem

• Multi-Camera Tracking
• Language-Guided Tracking
• Identity Association
• Uncertainty Propagation
• Camera Handoff

Innovation

• Recursive Event World Model
• Language-Guided Multi-Camera Tracking
• Association Uncertainty
• Structured Posterior Update
• Recurrent Belief