VLM-AutoDrive: Post-Training Vision-Language Models for Safety-Critical Autonomous Driving Events

In three linesVLM-AutoDrive is a post-training framework for adapting Vision-Language Models to safety-critical anomaly detection in autonomous driving. Fine-tuning on Nexar dashcam videos improves collision F1 from 0.00 to 0.69 and overall accuracy from 35.35% to 77.27% versus NVIDIA Cosmos-Reason1 7B zero-shot.

Summary generated by Claude — human-verified