SAVeD: A First-Person Social Media Video Dataset for ADAS-equipped vehicle Near-Miss and Crash Event Analyses

Contribution

A multimodal dataset project for first-person ADAS near-miss and crash event understanding.

This work anchors the multimodal traffic video and ADAS safety theme. It supports event parsing, near-miss/crash understanding, and future counterfactual safety analysis.

This dataset paper is separate from the SAVeD deployment-oriented warning protocol, which is listed as a submitted manuscript.