Outerview/road-safety-vulnerable-road-users

Road Safety & Vulnerable Road Users Visual Dataset

Rows: 484

Dataset Description

Road Safety & Vulnerable Road Users Visual Dataset is a Global street-level imagery and geospatial computer vision dataset for image classification, visual search, and mobility research. The labeled features in the dataset are Crosswalk, Cyclist, School Bus, Traffic Cone, and Wheelchair, and each row records a matched visual feature label derived from the Outerview internal visual search pipeline. The packaged files include data.csv, data.geojson, data.parquet, and a representative image archive images-batch-0001.zip for sample review. It is useful for traffic safety analysis, accessibility auditing, municipal asset review, AV perception benchmarking, insurance risk analysis, environmental analysis, and retail site selection workflows.

This dataset is a sample from Outerview's full visual index of 10M+ real-world observations, powered by 35B visuals and observations. Search the full dataset for free at outerview.ai. API access and full dataset available at outerview.ai. Developer docs at docs.outerview.ai.

When image archives are included, the packaged images are representative examples and larger-scale image generation, retrieval, and review should be done through the API.

Updated on a bimonthly basis.

Dataset Structure

Features in the dataset

• Crosswalk: crosswalk
• Cyclist: cyclist
• School Bus: school bus
• Traffic Cone: traffic cone
• Wheelchair: wheelchair

Schema

Column	Type	Description
filename	string	Image filename in the exported image batch.
image_folder	string	Relative folder inside the image zip batch.
date	string	Capture date or source timestamp when available.
country	string	Country associated with the visual record.
state	string	State or province associated with the visual record.
city	string	City or locality associated with the visual record.
feature	string	Matched visual feature label.
latitude	float64	Latitude in WGS84 decimal degrees.
longitude	float64	Longitude in WGS84 decimal degrees.
source	string	Source attribution for the visual record.

Dataset Creation

Source

Records are sampled from Outerview's global visual index using the Outerview internal visual search pipeline and query-driven embedding match over street-level and related imagery. The result is a compact sample of observations aligned to the requested road safety and vulnerable road user queries.

Annotation

Labels were assigned at the feature level only for Crosswalk, Cyclist, School Bus, Traffic Cone, and Wheelchair. No bounding boxes are included, and each record stores the matched visual feature label for image classification and retrieval use cases.

Uses

Intended Uses

Municipal asset teams can use this dataset for roadway condition review, crosswalk auditing, and accessibility planning. AV perception engineers can use it for image classification and scene understanding tasks related to vulnerable road users and traffic infrastructure. Insurance underwriters, environmental analysts, and retail site selection teams can use the geospatial metadata for visual search and location-based assessment.

Out-of-Scope Uses

Not intended for facial recognition, individual identification, or surveillance of private individuals.

License

This dataset is released under the CC BY 4.0 license. Free for research and commercial use with attribution.

Citation

@dataset{outerview_road-safety-vulnerable-road-users_2026,
  title={Road Safety & Vulnerable Road Users Visual Dataset},
  author={Outerview},
  year={2026},
  url={https://huggingface.co/datasets/outerview/road-safety-vulnerable-road-users}
}

About Outerview

Outerview is a research lab focused on understanding the physical world at planetary scale.

We are building systems that can organize the world's physical information and make it accessible and usable transforming raw imagery, video, location, and spatial context into knowledge that people and machines can search, interpret, and act on.

Our belief is simple: the physical world should be as searchable and understandable as the digital world.