01 · Mapping companies
Backlogs of imagery, inconsistent labels, and GIS outputs that need rework before they're usable.
You get schema-conformant feature layers, defined classes, QA notes, and agreed export formats for downstream review.
Geospatial annotation · documented GIS QA
Human-Verified Geospatial Annotation and GIS QA
For teams with captured imagery, point clouds, model proposals, or draft asset data that still needs qualified review, schema conformance, and an inspectable GIS delivery package.
38.9072°N, 77.0369°WSRID: 4326
Who it's for
Built for teams drowning in imagery.
Three kinds of teams hire us. If you recognize your own workflow here, we are probably a fit.
02 · Mobile-mapping primes
Street-level, LiDAR, and video capture piling up faster than you can extract from it.
Supported roadway assets are reviewed against source evidence, with spatial checks and exception flags tied to the relevant record or location.
03 · Infrastructure & DOT teams
Asset inventories and safety reviews stall the moment imagery becomes manual review labor.
You get an inspectable inventory package for supported assets, with limitations separated from observations that require field or authoritative-source verification.
Scope and controls
From source evidence to an acceptance-ready GIS package
Image-space labels identify what appears in a frame; real-world GIS coordinates require positioning evidence, a defined CRS, and validation. Attributes such as underground ownership, reflectivity, or field condition are not inferred when the source cannot support them.
Accepted inputs
- • Street-level or aerial imagery
- • LiDAR or point-cloud views
- • Model proposals or existing annotations
- • Draft GIS layers and asset records
Supported visible assets
- • Roadway signs and pavement markings
- • Hydrants, manholes, poles, guardrails, and curb ramps
- • Trees or vegetation when the source and requested taxonomy support observation
- • Other infrastructure features confirmed during scoping
GIS QA controls
- • CRS and datum handling
- • Geometry type, position, and topology
- • Required attributes and ontology
- • Duplicate, omission, provenance, exception, and acceptance-sample review
Acceptance package
- • Agreed GIS or annotation files
- • Schema or data dictionary
- • QA summary and exception list
- • Provenance, limitations, and acceptance notes
Automation with focused human review
Repeatable candidate generation, format checks, and rule-based validation can be automated. Qualified reviewers concentrate on ambiguous geometry, consequential attributes, taxonomy conflicts, omissions, duplicates, and exceptions that require judgment.
Inspect the annotation and GIS QA demonstrationDelivery sample
The exact format your team receives.
A clearly labeled demonstration of annotated frames, a representative feature record, and a QA trail. It illustrates the delivery structure; it is not a client result or production-accuracy claim.
DEMO FRAME_00842.JPG · REVIEWED
Ladder crosswalk
Image observation linked to a representative GIS record; field-only condition attributes are not inferred.
DEMO FRAME_00843.JPG · REVIEWED
Fire hydrant
Visible asset classification with a review flag where real-world position or ownership needs another source.
DEMO FRAME_00847.JPG · REVIEWED
Regulatory stop sign
Visible class reviewed; condition and reflectivity require agreed evidence and are not assumed from this frame.
delivery.geojsonSAMPLE
{
"type": "Feature",
"geometry": {
"type": "Point",
"coordinates": [-77.03687, 38.90719]
},
"properties": {
"asset_id": "SIGN_2026_04128",
"class": "regulatory_stop",
"mutcd_code": "R1-1",
"qa_status": "reviewed",
"exception_code": null,
"capture_date": "2026-05-14",
"crs": "EPSG:4326"
}
}Also delivered as: GeoPackage · Shapefile · FileGDB · CSV · PostGIS · COCO · KITTI · Mapillary · custom schema
Want to see these deliverables in action? Explore five interactive proof pages — source imagery, map-linked assets, QA decisions, and export outputs.
See the demo path →Demo path
See the annotation work from raw imagery to GIS delivery.
Proof pages that follow the buyer path after the annotation service page: feature review, QA, lineage, and the final files a GIS team can inspect before approving a pilot.
Delivery workflow
How a pilot works.
We start small. Once the schema and QA results are clear, the same rules scale across your full inventory.
01
Send sample data
You share imagery, video frames, LiDAR views, or an existing GIS layer with the assets you need extracted.
02
Lock schema and QA rules
Our team confirms classes, geometry, attributes, acceptance criteria, and delivery format before production starts.
03
Annotate and review
Our team labels features, validates geometry, enriches attributes, and flags uncertain cases through multi-stage QA.
04
Deliver GIS-ready output
You receive files, QA notes, an exception log, and a handoff package ready for mapping, AI, or asset workflows.
→Pilot timing and acceptance criteria are confirmed after the sample, schema, and review rules are understood.
Pilot and delivery scope
Agree on acceptance before production begins.
Start with a bounded sample, lock the schema and review rules, then choose a follow-on delivery model only after both teams inspect the result.
Pilot
Sample pilot
Scopedto the sample
Provide representative source data, target classes, required attributes, coordinate requirements, and acceptance rules. GSS returns an agreed sample package for review.
Request a pilotOngoing capacity
Follow-on delivery
Definedafter pilot review
Batching, review depth, turnaround, change control, and reporting are documented from the accepted pilot rather than promised before the data is examined.
Request a pilot+−See scope controls
- Batching
- Set from representative volume, asset mix, and exception rate
- Turnaround
- Confirmed after sample review and workflow calibration
- Change control
- Schema, taxonomy, and acceptance-rule changes are documented
Acceptance criteria
Define geometry, attribute, taxonomy, duplicate, omission, and sampling rules before production.
Inspectable exceptions
Ambiguous or unsupported observations remain visible in an exception list rather than being silently guessed.
Delivery evidence
The handoff can include a schema or data dictionary, QA summary, provenance, limitations, and agreed output files.
Change control
Material changes to sources, ontology, or acceptance rules are reviewed before they alter the production workflow.
Start with a pilot.
Send us your data type and volume. We'll review the details and respond with the next step for scoping.