Multi-camera portal
Fixed camera array capturing every container face under consistent geometry and lighting as it passes the gate.
AI intelligence system
CargoInspect
From inspection image to defensible damage evidence.
An automated container and cargo inspection system that captures condition at entry and exit, classifies damage consistently, and produces an evidence package that stands up in a claim.
- Sector
- Logistics, Freight & Insurance
- Year
- 2024
- Duration
- 6 months
- Team
- 6 people
- Platforms
- Gate kiosk · Web · Mobile
- Status
- In production
6-face
Capture coverage
Every face, controlled geometry
Entry vs exit
Comparison basis
Custody delta, not opinion
Consistent
Severity grading
Independent of who is on shift
Claim-ready
Evidence package
Generated automatically
01 — The problem
Damage disputes were settled by argument. A container arrived damaged, the terminal said it arrived that way, the carrier said it did not, and the evidence was a handful of phone photographs taken from inconsistent angles by whoever was on the gate.
The inconsistency was the real cost. Two inspectors graded the same dent differently, so claims outcomes depended on who was working that shift rather than on the condition of the box.
02 — What we did
We standardised capture before touching classification. A fixed multi-camera portal photographs every face of every container under controlled geometry and lighting, so entry and exit images are directly comparable rather than approximately similar.
With comparable capture, difference becomes provable. The system aligns entry and exit imagery, isolates what changed during custody, and produces a package with both images, the delta, the classification and the full chain of custody attached.
You cannot prove a difference between two photographs taken from different angles.
Interface
What it looks like in use
Two of the screens that carry the most weight in daily use, rebuilt here from the production design system.
MSKU 447188-2DELTA FOUND
Entry · 14 Mar 06:12 · grade B6 FACES
Exit · 19 Mar 17:48 · grade C6 FACES
Left panel · new deformation 340mmDURING CUSTODY
Repair estimate · code DP-3$1,140
Claim package generated
Difference becomes provableControlled capture geometry makes entry and exit images directly comparable, so what changed during custody is isolated rather than argued.
Damage by handler · 90 daysCONSISTENT GRADING
Terminal 3 · door damage3.1x MEAN
Route AE-NL · corrosion1.2x MEAN
Findings per 1,000 moves
Patterns across the fleetConsistent grading made damage concentration by route and handler visible for the first time.
Capabilities
What the system does
Grouped by the job each set of capabilities exists to do, rather than by which team built it.
Capture
01 · Controlled by designIdentity capture
Container number and seal read automatically and reconciled against the booking.
Quality gating
Occlusion, motion blur and exposure problems detected at capture so the pass can be repeated immediately.
Mobile supplement
Inspectors add close-range detail shots that inherit the same metadata and chain of custody.
Assessment
02 · Consistent gradingDamage detection and segmentation
Dents, punctures, corrosion, door damage and structural deformation localised on the container surface model.
Severity assessment
Graded against the industry damage taxonomy so results are consistent across shifts and terminals.
Before and after comparison
Entry and exit imagery aligned to isolate exactly what changed during custody.
Repair estimation
Detected damage mapped onto standard repair codes and estimated cost.
Evidence and claims
03 · The commercial payoffReport generation
Inspection reports produced automatically with annotated imagery and classification rationale.
Insurance evidence package
A complete bundle: both captures, the delta, grading, timestamps and custody chain.
Dispute support
Side-by-side presentation designed for a claims adjuster rather than an engineer.
Fleet condition trends
Damage patterns tracked by route, handler and equipment type.
Architecture
How it is put together
Layer by layer, with the reason each one exists — because the reason is usually the interesting part.
01Capture portal
Synchronised multi-camera rig with controlled lighting and trigger detection, producing a consistent capture set per pass.
- Industrial cameras
- Hardware trigger
- Controlled lighting
02Identity resolution
OCR of container and seal numbers reconciled against terminal booking systems.
- OCR
- Terminal integration
03Damage models
Segmentation and classification trained against the industry damage taxonomy with calibrated severity output.
- PyTorch
- Segmentation
- Taxonomy mapping
04Comparison
Geometric alignment of entry and exit captures with change isolation on the container surface model.
- Image registration
- Change detection
05Evidence service
Immutable storage of captures and findings with signed, exportable evidence bundles.
- S3 object lock
- Signed exports
Technology
Vision
- PyTorch
- Segmentation
- Image registration
- OCR
Hardware
- Industrial cameras
- Edge compute
- Controlled lighting
Backend
- Python
- FastAPI
- PostgreSQL
- S3
Frontend
- React
- TypeScript
- Annotation tooling
Outcome
What changed
Measured against how the operation ran before, not against a benchmark chosen after the fact.
-
Grading stopped depending on the shift. The same damage receives the same severity regardless of who is on the gate.
-
Custody deltas are provable. Controlled capture geometry makes entry-to-exit comparison meaningful.
-
Claims packages are automatic. Evidence bundles are generated from the inspection record, not assembled after a dispute.
-
Patterns emerged across the fleet. Damage concentrated by route and handler became visible for the first time.
How it ran
-
Months 1-2
Capture design
Designed and validated the portal geometry, because comparability depends entirely on it.
-
Month 3
Identity and integration
Container and seal OCR with terminal system reconciliation.
-
Months 4-5
Damage models
Detection, taxonomy-aligned grading and entry-exit change isolation.
-
Month 6
Evidence and rollout
Report generation, signed evidence bundles and gate deployment.
Related work
From the same practice
Manufacturing & Production Quality
LineTrace
From defect detection to defect causality.
A multimodal system that correlates line video, machine telemetry, batch records and operator actions to explain why a defect happened, not merely that it did..
2025 · 10 months
Telecommunications & Network Operations
NetCause
Not alarm detection. Failure hypothesis reduction.
A correlation engine that turns an alarm storm into a short ranked list of probable root causes by reasoning over network topology and fault propagation rather than alarm counts..
2025 · 8 months
Oil & Gas · Asset Integrity
PipeIntegrity
From a single inspection to longitudinal asset integrity.
A vision system that detects and localises pipeline deterioration, then tracks each defect across inspection cycles so maintenance is prioritised by progression rather than by appearance..
2024 · 9 months
Something similar?
Running into the same problem CargoInspect solved?
Tell us what you run. We will reply within two business days with what we would build for your situation — and, just as usefully, what we would leave out.