Hard Intersection Multimodal Samples is a curated multimodal dataset of accident-prone urban intersections in Japan for autonomous driving research. It provides multi-camera images, trajectory data, HD maps, semantic annotations, point cloud data, and 3DGS assets. The dataset was created by dynamic-maps and was last updated on June 10, 2026.
Use Cases
- Train perception models for autonomous vehicles based on multi-camera images and semantic annotations.
- Develop and validate trajectory prediction algorithms using the provided trajectory data.
- Generate synthetic environments for simulation testing using HD maps and 3DGS assets.
- Analyze accident-prone scenarios in urban intersections using the multimodal sensor data.
Strengths
- Dataset is curated specifically for accident-prone urban intersections.
- Provides multiple data modalities including images, trajectories, maps, annotations, and point clouds.
- Data was captured via an industrial-grade Mobile Mapping System (MMS).
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count, file formats, and dataset size are unknown, which may limit suitability assessment.
- The dataset is a sample; its full scope and completeness are unclear.
Provenance
- Source
- dynamic-maps
- Collection Method
- Captured via an industrial-grade Mobile Mapping System (MMS) on public roads.
- Freshness
- Last updated 2026-06-10 11:28:06; freshness should be verified.
- Geography
- Japan, specifically accident-prone urban intersections.