Platform
Operator value
Why railway teams care.
Teams can move from disconnected imagery to structured vegetation events, validation status, and location-based maintenance planning.
Software and Data Solution
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.
Documented basis
The DriveTrust project implemented a production-grade vegetation pipeline with human-in-the-loop validation and retraining on real Portuguese rail network data.
Vegetation workflow
From corridor imagery to validated vegetation events.
RailEyes turns captured corridor imagery into reviewable vegetation events by combining rail-trained segmentation, zone-aware filtering, depth context where appropriate, and operator validation. Corrected findings can feed retraining so the model improves around real railway conditions.
Rail-specific vegetation classes for corridor review.
Zone and distance context to reduce review noise.
Human validation for operational confidence and model improvement.
Investor value
Why this can compound.
Validated rail-specific vegetation data can become a defensible dataset for infrastructure inspection, seasonal planning, and future predictive maintenance.
Workflow
How RailEyes Vegetation Detection fits into the RailEyes platform.
Capture forward-facing corridor imagery from operating trains.
Run segmentation to identify vegetation classes in the scene.
Filter detections by zones and depth where appropriate.
Let reviewers mark events as correct or wrong for model improvement.
Outputs
What the module produces.
Vegetation eventsValidation workflowRail location contextRetraining data