Traffic data that survives an audit.
CORE AI replaces manual roadside surveys with continuous video analysis. It measures every vehicle, reconstructs true origin-destination movement, and retains a verifiable detection log for each result.
- ±2%
- Count error margin
- 24/7
- Continuous capture
- 100%
- Measured, not modeled
Manual surveys were never built to be defensible.
Infrastructure decisions worth hundreds of millions rest on traffic figures collected by hand over a few hours. Four structural weaknesses recur in almost every study.
Human error in manual counting
Enumerators fatigue, miscount at peak volume, and disagree between sites. The error is silent and unquantified in the final report.
Origin-destination data is modeled, not measured
Conventional studies infer movement from gravity models and roadside interviews. The resulting matrix is an assumption presented as evidence.
Incomplete time coverage
Short field surveys sample a fraction of a day and are extrapolated across weeks. Off-peak, night, and seasonal behaviour go unobserved.
No auditable evidence trail
A tally sheet cannot be re-examined. When a figure is challenged during due diligence, there is nothing underneath it to verify.
A four-layer pipeline from raw video to structured data.
Each layer transforms the output of the last. Nothing is estimated; every figure in the final matrix traces back to an individual detection.
The same five questions, answered two ways.
Measurement duration
Manual survey
4–12 hours, sampled
CORE AI
Continuous, 24/7
Error margin
Manual survey
±15–20%, unquantified
CORE AI
±2%, reported per site
Origin-destination source
Manual survey
Modeled & interviewed
CORE AI
Measured from tracked movement
Auditability
Manual survey
Tally sheets, not reviewable
CORE AI
Full detection log per figure
Field personnel required
Manual survey
Large enumerator teams
CORE AI
None after installation
Six structured outputs, ready for your models.
Each engagement delivers the same defined set of datasets and reports, in formats that drop directly into transport modeling and GIS environments.
Traffic intensity reports
Volume by direction, hour, and day across the full capture window – not a peak-hour extrapolation.
Origin-destination matrices
Movement between entry and exit points, reconstructed from tracked vehicles rather than modeled demand.
Travel time analysis
Point-to-point journey times across the network, distributed by time of day and vehicle class.
Vehicle classification data
Counts broken down by passenger car, light goods, heavy goods, and bus for every counting location.
Transit vs. local split
The share of through-traffic against locally originating movement – critical for bypass and corridor cases.
Audit-ready detection log
The underlying record of every detection with timestamp, class, and confidence, available for third-party review.
Deployed on an active feasibility study.
CORE AI is currently supporting a technical assistance engagement with a multilateral development bank on an urban transport feasibility study in Eastern Europe.
Origin-destination data for the study was measured in full – no manual roadside interviews were used at any station.
Built for institutional audit requirements.
The system is designed around the constraints that govern publicly financed infrastructure work: data protection, verifiability, and integration with existing institutional tooling.
Local, closed-environment processing
Video is processed on-site within a closed environment. No footage is transmitted to external or cloud servers at any stage of the pipeline.
Cryptographically hashed identifiers
Vehicle identifiers are hashed at the point of detection. Every output contains only irreversible hashes – no readable plates leave the system.
Third-party reviewable logs
The complete detection log is retained and can be released for independent verification, supporting donor compliance and due-diligence review.
Open API integration
Structured outputs are exposed through an open API for direct integration with GIS platforms and standard transport modeling software.
Three roles, one requirement: data that holds up.
Infrastructure consultants
Replace subcontracted manual counts with a measured dataset you can stand behind when a client or reviewer questions the numbers.
Government transport agencies
Support capital allocation with measured demand rather than estimated figures, and retain the evidence for public scrutiny.
International financing institutions
Fund projects on traffic data that passes donor compliance and independent due-diligence review without qualification.
A defined four-week engagement.
Fixed scope, fixed timeline, lump-sum pricing. Payment falls due on delivery and acceptance – no advance payment is required.
Setup & calibration
Camera stations installed and calibrated against site geometry. Detection thresholds tuned to local conditions.
Continuous processing
Uninterrupted capture and analysis begins across all stations, running around the clock.
Continuous processing
Capture continues to build the full dataset across weekdays and weekend behaviour.
Full delivery
All six deliverables and the detection log are compiled, validated, and handed over.
Traffic evidence you can put in front of an auditor.
If your study needs measured origin-destination data with a verifiable record behind every figure, we would welcome a technical conversation.