Platform
01
Capture from scheduled trains
Retrofit camera units collect visual, GPS, thumbnail, device, and motion context during normal railway operation.
AI-assisted railway inspection
Turn scheduled rail operations into infrastructure intelligence.
RailEyes helps railway operators capture corridor evidence from active rolling stock, anonymize sensitive imagery, detect inspection-relevant conditions, and review findings on a rail-aware map.
Infrastructure intelligence
From ordinary train movements to inspection-grade evidence.
As RailEyes data accumulates, the corridor view becomes more than video: it becomes a structured evidence layer for maintenance, vegetation, asset, and safety teams.
Scheduled capacity, inspection value
Turn trains that already run into a repeatable inspection network.
RailEyes helps operators use scheduled rail capacity to collect corridor evidence more often, reduce the need for separate inspection runs, prioritize maintenance with clearer context, and compound value through reusable rail-specific data.
Want to learn more?Report-backed proof points
Field-validated railway AI, built from real operating constraints.
RailEyes was developed through the DriveTrust initiative with CP and IP, moving from concept to retrofit hardware, edge capture, anonymization, vegetation analytics, and map-based review workflows on active Portuguese rolling stock.
18 months
Pilot and monitoring period
The project expanded from an initial feasibility window into a longer validation program as railway, privacy, hardware, and data access requirements were worked through.
2 train families
Retrofit tested
RailEyes camera units were deployed on UTE 2240 and CPA 4000 series rolling stock, with hardware iterations shaped by vibration, temperature, visibility, and installation constraints.
90%+
Lower data footprint in TrackView concept
Project reporting describes replacing full-length video review with anonymized still frames every 50-100 meters for targeted corridor inspection.
Rail-specific datasets
Human-validated inspection data can improve models over time and deepen defensibility.
Privacy-first European adoption
Anonymization and selective media handling reduce friction for public infrastructure environments.
Existing rolling stock as capture network
Retrofit deployment can turn scheduled train operations into recurring inspection coverage.
Railway maintenance
Inspection evidence built around real railway operations.
RailEyes is designed for infrastructure teams that need useful data without adding a dedicated inspection vehicle for every review cycle. The platform combines retrofit capture hardware, edge-side data handling, cloud analytics, and human validation into a practical inspection workflow.
Operator value
Use routine train movement to collect visual evidence, prioritize field work, and create a clearer record of vegetation, assets, and trackside conditions.
Investor value
The RailEyes platform compounds value through rail-specific datasets, reusable mapmatching, privacy-oriented media pipelines, and modular analytics that can expand across inspection categories.
Platform
A railway inspection stack with defensible layers.
The strongest RailEyes story is not one AI model. It is the combination of capture hardware, selective upload, anonymization, rail location intelligence, validation loops, and operator-facing review tools.
02
Reduce data before it becomes expensive
Metadata and thumbnails can move first, while full media is uploaded selectively for segments of interest or deeper review.
03
Anonymize and structure evidence
Captured images and video can be anonymized, matched to rail locations, and routed into vegetation, zoning, TrackView, or asset review workflows.
04
Improve with validation
Human review of AI findings creates a feedback loop for better rail-specific models and clearer operator confidence.
Rail-trained perception
Semantic segmentation enriched with depth context and validation data from real railway corridors.
Capabilities
Products that map to concrete railway workflows.
RailEyes keeps inspection content specific: vegetation monitoring, TrackView corridor review, rail geolocation, anonymization, zoning, motion context, and aggregated infrastructure intelligence.
Hardware and Edge Intelligence
Software and Data Analytics
RailEyes TrackView
Map-based track review using anonymized still frames extracted from onboard video at distance or time intervals, so teams can inspect long corridors without moving full video around by default.
Read more
Real-Time Railway Stream
Live and near-real-time railway corridor streaming for operators who need immediate visual context from scheduled train movement, existing camera feeds, or RailEyes capture units.
Read more
RailEyes Zoning
Location-aware object and obstacle review that separates trackside observations into operational zones such as LEFT, RIGHT, and DANGER.
Read more
RailEyes Vegetation Detection
Railway vegetation intelligence using semantic segmentation for corridor growth, including high grass, low grass, tree trunks, tree tops, and miscellaneous vegetation classes.
Read more
RailEyes Map/Track/Lane Matching
Rail-specific location matching that connects GPS observations to tracks, segments, stations, and kilometer points instead of relying on road-oriented map matching.
Read more
RailEyes Anonymization
Privacy-oriented video and image anonymization for railway review workflows where pedestrians, passengers, staff, or public spaces may appear in captured data.
Read more
RailEyes Dynamics
RailEyes Dynamics blends onboard inertial signals, visual corridor proof, and rail-specific map matching so abnormal motion patterns can be reviewed at the correct track location.
Read more
RailEyes Mapmatching
Hectometer-oriented matching of railway observations using GPS, motion context, and rail network structure for more actionable inspection records.
Read more
RailEyes Heatmaps
Aggregated visual intelligence for detecting repeated vegetation, asset, obstacle, and passenger-presence patterns across corridors and stations.
Read more
How it works
From corridor capture to reviewed findings.
RailEyes helps teams understand what was captured, where it belongs on the rail network, whether sensitive imagery has been handled correctly, and what deserves maintenance attention.
Capture visual, GPS, thumbnail, and motion data from operating trains.
Upload automatically where useful and request richer media only when needed.
Run AI-assisted detection for vegetation, objects, zones, and review concepts.
Validate findings through a cloud workflow so outputs stay useful to maintenance teams.
Deployment contexts
Built for corridors, dense urban rail, and repeatable transit routes.
The same RailEyes architecture can be configured around different rail environments, from long main-line corridors to dense station-adjacent networks.
Main Line Rail
Long-distance railway corridor monitoring for vegetation, trackside assets, and periodic visual review using scheduled train movement as the capture opportunity.
Urban Rail
Dense railway environments where trackside clearance, human presence, stations, and public-space privacy require structured evidence and careful anonymization.
Metro and Dedicated Transit
Repeatable route monitoring for dedicated-grade networks where frequent service, tight clearances, and recurring inspection cycles can benefit from consistent capture.
Detection examples
Visual inspection evidence for trackside review.
These examples show the kind of railway imagery and AI-assisted overlays that support operator review. Production claims should be tied to the specific corridor, pilot scope, and validation method agreed with each operator.
