From Street Imagery To Infrastructure Insights

Source: Product Overview

Source: Access And Setup

Start With Evidence. Define The Question.

Define the area, select the imagery variant, and add LiDAR when the question requires 3D geometry.

Source: Street View Insights × Telco — Slide 15

One Asset Can Have Multiple Observations.

For the base pole and sign tables, keep the physical object, its recorded appearances, and the dataset version distinct. Customer detections from panoramic imagery need their own data model.

Source: Schema Reference

Turn A Question Into A Repeatable Analysis.

Google provides sample notebooks for exploring data, classifying signs, and analyzing utility poles in Vertex AI Colab Enterprise.

Sample notebooks demonstrate techniques. They do not establish a guaranteed accuracy level for your geography or assets.

Source: Official Sample Notebooks

Choose The View That Answers The Question.

Cropped imagery isolates an asset. Full-frame imagery adds its surroundings. Panoramic data supports customer-defined detection across a scene.

Panoramic data supports work on assets such as cabinets, driveways, and continuous infrastructure. It does not inherit the ready-made pole and sign locations supplied with cropped and full-frame variants. LiDAR adds 3D geometry aligned with the selected imagery variant.

Source: Street View Insights × Telco — Slide 15

Add Geometry To What You Can See.

Combine imagery with a 3D point cloud to measure spatial relationships and locate assets detected by your own models.

From Shape To Measurement

Select the asset and relevant points, establish the local coordinate frame, and calculate the dimension needed for the decision.

• Pole height and attachment positions.
• Distances between equipment, wires, and surrounding objects.
• Vegetation proximity and candidate clearance issues.
• Geometry for infrastructure models and digital twins.

Coordinate storage precision does not establish measurement accuracy. Validate point selection, alignment, and the measurement method against trusted reference dimensions.

Source: Street View Insights × Telco — Slide 10, 15, 17

Technical basis: supplied LiDAR sample specification v0.9 · PROJ: ECEF To Local Coordinates

What Is In The Sample Data?

The supplied v0.9 specification describes point clouds, photographs, and camera calibration. These details describe the sample export rather than a universal data contract.

Source: user-supplied Google Maps Platform LiDAR Product Specifications (Sample Data), v0.9. No point-cloud files or classification dictionary were supplied for verification.

Connect The Data To Field Decisions.

Select an imagery variant and an analysis method around the decision your team needs to make.

Source: Street View Insights × Telco — Slide 7–11, 17, 32–33

A Missing Detection Is Not A Missing Asset.

Use these interpretation checks when turning imagery analysis into an operational recommendation.

These are recommended interpretation controls for a pilot. Use field records and qualified review where the decision requires them.

Source: Regional And Temporal Coverage

Source: Capture And Snapshot Metadata

Count Recorded Signs Before Assessing Condition.

Begin with a reproducible inventory question: how many distinct road-sign assets are represented in the latest snapshot of this subscribed dataset?

-- Illustrative GoogleSQL; replace the project and dataset.
-- Counts distinct road-sign assets in the latest snapshot.
SELECT
  snapshot_id,
  COUNT(DISTINCT asset_id) AS road_sign_assets
FROM `YOUR_PROJECT.YOUR_LINKED_DATASET.latest_assets`
WHERE asset_type = 'ASSET_CLASS_ROAD_SIGN'
GROUP BY snapshot_id
ORDER BY snapshot_id;

Next, scope the area, review capture dates in observations, and evaluate a condition-classification workflow. Keep its labels separate from the inventory count. The table path and output above are placeholders.

Source: Table And Asset-Type Definitions

Onboard, Subscribe, Then Inspect.

Access begins with a request to Google. An API key alone does not complete the documented onboarding workflow.

US/EU here describes data consumption regions, not a promise of imagery coverage across those territories.

Source: Onboarding Instructions

Source: Usage Policies

Budget For Data Access And Cloud Analysis.

Google directs pricing, contracting, and account questions to your Google representative. Obtain a scoped quote before building a cost case.

No per-image price or free allowance is assumed here. Estimate a bounded sample first, record actual consumption, then model the larger workload.

Source: Pricing And Cloud Support

Make The Pilot Answer One Useful Question.

Suggested pilot: determine whether imagery analysis can help a road agency prioritize signs for manual review.

Bring An Area, An Asset Class, And A Decision.

Start with a specific question and the records you would use to verify the answer. Then confirm access and test a small, documented workflow.

Sources: official documentation, the supplied Street View Insights × Telco deck, and LiDAR sample specification v0.9. Examples describe workflows; they do not report an executed customer analysis.