VLAlert-Bench: A Unified Benchmark for Driving-Alert Decision Making
Overview
VLAlert-Bench integrates six driving-event datasets into a single per-tick prediction task with three actions — SILENT, OBSERVE, and ALERT. At each one-second tick a model sees the last 8 frames and predicts the appropriate alert action. The benchmark provides 13,534 videos across five splits, 192,892 one-second labeled ticks, per-frame action labels, and split manifests, and hosts the full ADAS-TO-Critic mp4 corpus directly. Source datasets include Nexar Collision, DoTA, DAD, DADA-2000, ADAS-TO-Critic, and the Kaggle Accident set. Released under CC-BY-4.0.
Model architecture for per-tick alert decisions.
Belief spans over the SILENT / OBSERVE / ALERT states.
Label distribution across the validation splits.
Demos
Per-tick alert decisions on real driving-event footage.
Alert decision on an approaching hazard.
Escalation from OBSERVE to ALERT.
A safety-critical collision sequence.
BibTeX
@misc{wang2026vlalert,
title = {VLAlert-Bench: A Unified Benchmark for Driving-Alert Decision Making},
author = {Wang, Yuhang and Zhou, Hao},
year = {2026},
howpublished = {Hugging Face Datasets},
url = {https://huggingface.co/datasets/HenryYHW/VLAlert}
}