CADD: A Chinese Traffic Accident Dataset for Statute-Based Liability Attribution

As autonomous driving technology advances, the critical challenge evolves beyond collision avoidance to the $\mathbf{\text{adjudication}}\mathbf{\text{of}}\mathbf{\text{liability}}$ when accidents occur. Existing datasets, focused on detection and localization, lack the annotations required for this legal reasoning. To bridge this gap, we introduce the $\mathbf{\text{C}}$hinese $\mathbf{\text{A}}$ccident $\mathbf{\text{D}}$uty-determination $\mathbf{\text{D}}$ataset ($\mathbf{\text{CADD}}$), the first benchmark for statute-based liability attribution. CADD contains 792 real-world driving recorder videos, each annotated within a novel $\mathbf{\text{“Behavior–Liability–Statute”}}$ pipeline. This framework provides $\mathbf{\text{granular,}}\mathbf{\text{symmetric}}\mathbf{\text{behavior}}\mathbf{\text{annotations}}$, clear responsibility assignments, and, uniquely, links each case to the specific $\mathbf{\text{Chinese}}\mathbf{\text{traffic}}\mathbf{\text{law}}\mathbf{\text{statute}}$ violated. We demonstrate the utility of CADD through detailed analysis and establish benchmarks for liability prediction and explainable decision-making. By directly connecting perceptual data to legal consequences, CADD provides a foundational resource for developing accountable and legally-grounded autonomous systems.