Road segments

The Road segment type (RoadSegmentDatasetDTO) represents an individual segment of an OSM road, with detailed parking statistics for that precise segment.

Description

Unlike Roads, which aggregate data across an entire way, segments enable fine-grained analysis of parking. Each OSM road is split into segments of homogeneous length, making it possible to visualize density variations along a single street.

Relationship with Roads

OSM road "Rue des Lilas" (osm_id: way/123456789)
│
├── Segment 0 (id: way/123456789_0)
│   └── 5 spots catalogued
│
├── Segment 1 (id: way/123456789_1)
│   └── 12 spots catalogued
│
├── Segment 2 (id: way/123456789_2)
│   └── 0 spots (no-parking area)
│
└── Segment 3 (id: way/123456789_3)
    └── 8 spots catalogued

Properties

Identification

Property	Type	Description	Example
id	string	Unique identifier ({osm_id}_{segment_index})	way/123456789_2
layer_type	string	Document type	road_segment
osm_id	string	OSM identifier of the parent road	way/123456789
segment_index	integer	Index of the segment along the road	2
h3_index	string	H3 cell the segment is attached to	893b0c6a9a7ffff

Geometry

Property	Type	Description
geometry	LineString	Segment geometry (GeoJSON)

Statistics

Property	Type	Description
segment_length_meters	number	Segment length in meters
total_spots	integer	Number of spots catalogued on the segment
avg_fit_score	number	Average confidence score (0-1)

Traceability

Property	Type	Description
captured_at	datetime	Date of the most recent imagery for this segment

Use cases

Graduated visualization

Use total_spots to build a map with a color gradient:

Spot count	Suggested color	Interpretation
0	Grey	No parking
1-5	Light blue	Low density
5-10	Blue	Medium density
10-50	Dark blue	High density
50+	Very dark blue	Very high density

Identifying critical areas

Analyze segments to spot:

• Segments with no parking (no-parking areas, double-parking)
• Concentrations of spots (linear parking)
• Density variations along a street

Analysis by H3 cell

Cross-reference segments with H3 cells to:

• Compute parking density per zone
• Compare provision between neighborhoods
• Identify under-served areas

Recommended QGIS style

To visualize segments with a gradient based on total_spots:

1. Download the QGIS style file: road_segments.qml
2. In QGIS: right-click on the layer → Properties → Symbology
3. Style → Load Style → select the .qml file

Download and styles guide →

GeoJSON example

{
  "type": "Feature",
  "properties": {
    "id": "way/123456789_2",
    "layer_type": "road_segment",
    "osm_id": "way/123456789",
    "segment_index": 2,
    "h3_index": "893b0c6a9a7ffff",
    "segment_length_meters": 45.7,
    "total_spots": 8,
    "avg_fit_score": 0.82
  },
  "geometry": {
    "type": "LineString",
    "coordinates": [[55.47, -21.12], [55.475, -21.125]]
  }
}

Relationship with other types

Type	Relationship
Roads	A segment belongs to a road (via osm_id)
H3 cells	A segment belongs to an H3 cell (via h3_index)
Parking areas	Parking areas are located along segments

Variants by source

Road segments come in two variants depending on the imagery source:

Street-level segments (road_segments_street)

Computed from street-level imagery. Each 20 m segment carries a detailed left/right analysis.

Characteristic	Detail
Accuracy	~1-2 meters
Attributes	Left/right distinction, parking type per side
Search radius	10 meters around the segment

Satellite segments (road_segments_satellite)

Computed from HD satellite imagery. Aggregation is global per segment (no left/right distinction, since the nadir view does not allow for it).

Characteristic	Detail
Accuracy	~3-5 meters
Attributes	Total spots per segment only
Search radius	15 meters around the segment
Road filtering	Motorways, pedestrian ways and cycle paths excluded

Both variants share the same data format (RoadSegmentDatasetDTO), the same geometry (20-meter segments), and the same confidence scoring (avg_fit_score).