Contribution
Organizes vision-language traffic-safety research around event understanding, grounded evidence, and verification, with an agenda for generative and counterfactual analysis.
This review develops a framework for traffic video safety analysis. Its feedback process involves expert review and refinement; it is not an autonomous control loop or evidence of deployment readiness.
Publisher record. Published online in July 2026, assigned to the October 2026 issue.