Street-Level Imagery Dataset
nyuuzyou

@article{,
title= {Street-Level Imagery Dataset},
journal= {},
author= {nyuuzyou},
year= {},
url= {https://huggingface.co/datasets/nyuuzyou/streetview},
abstract= {# Street-Level Imagery Dataset

Metadata for street-level imagery across Eastern Europe and Northern Asia. Each record includes image URLs, coordinates, camera orientation, timestamps, and links to similar images captured at the same location over time.

## Summary

| Statistic | Value |
|-----------|-------|
| Total Records | 934,191 |
| Unique Images | 905,940 |
| Unique Sources | 819 |
| Time Span | 2016–2025 |
| File Format | Parquet (Zstd) |
| Compressed Size | 461 MB |

## Geographic Coverage

| Boundary | Value |
|----------|-------|
| Minimum Longitude | 20.49° E |
| Maximum Longitude | 152.32° E |
| Minimum Latitude | 38.55° N |
| Maximum Latitude | 69.05° N |

Coverage spans urban centers and rural routes. Density is higher in populated areas.

## Camera Specifications

**Directions**

| Direction | Count |
|-----------|-------|
| Front | 740,079 |
| Right | 194,112 |

**Resolutions**

| Preview Size | Full Size | Count |
|--------------|-----------|-------|
| 284×160 | 1920×1080 | 932,171 |
| 90×160 | 1080×1920 | 1,886 |
| 284×160 | 1536×864 | 77 |
| 213×160 | 2016×1512 | 41 |

## Data Structure

### Fields

| Field | Type | Description |
|-------|------|-------------|
| `id` | string | Unique image identifier |
| `sourceId` | string | Source device or session identifier |
| `heading` | float64 | Camera heading (0–360°) |
| `cameraDirection` | string | Mount position (`front` or `right`) |
| `timestamp` | string | ISO 8601 capture time |
| `imagePreview` | struct | Thumbnail URL and dimensions |
| `imageFull` | struct | Full resolution URL and dimensions |
| `pos` | array | [longitude, latitude] |
| `geometry` | struct | GeoJSON Point geometry |
| `similar` | array | Related images at location |
| `targetGeometry` | struct | Optional target reference |

### Image URL Schema

Two resolution variants per entry:

```json
{
  "imagePreview": {
    "url": "https://...",
    "width": 284,
    "height": 160
  },
  "imageFull": {
    "url": "https://...",
    "width": 1920,
    "height": 1080
  }
}
```

### Temporal Links

Records reference similar images from other timestamps at the same coordinates. Average of 14.3 links per location.

## Limitations

Image URLs may [rot](https://en.wikipedia.org/wiki/Link_rot). Coverage concentrates in urban areas, and historical density varies by location.

## License

Research use permitted. Comply with source terms of service and local data regulations.},
keywords= {},
terms= {},
license= {},
superseded= {}
}