AI that solves real problems

Narrative, benefits and goals for each technical application.

☁️

Point-cloud segmentation

On sites with billions of points, separating terrain, structure and vegetation manually takes weeks. With ML, a trained model classifies every point in seconds.

More than 90% time savedConsistent, reproducible resultsScales across multiple campaigns

  • More than 90% time saved
  • Consistent, reproducible results
  • Scales across multiple campaigns
🏺

Ceramic classification

During an excavation, cataloguing ceramic fragments is repetitive work. A classifier trained on standardised photographs proposes typologies for the specialist to review.

Faster cataloguingLess fatigue-related human errorAn accumulating dataset that improves over time

  • Faster cataloguing
  • Less fatigue-related human error
  • An accumulating dataset that improves over time
📄

Report automation

Technical photogrammetry reports follow a standard pattern. A pipeline that reads project metadata and generates the document reduces the work to minutes.

Complete, standardised reportsNo copy and pasteFull traceability

  • Complete, standardised reports
  • No copy and paste
  • Full traceability
🔍

Heritage anomaly detection

Compare a monument's state between campaigns to spot deterioration with change-detection algorithms on point clouds or images.

Early deterioration alertsQuantitative monitoringSupport for conservation plans

  • Early deterioration alerts
  • Quantitative monitoring
  • Support for conservation plans
📐

3D feature extraction

Automatically identify building elements—arches, columns and mouldings—in complex 3D models for cataloguing and analysis.

Fast cataloguingComparative analysis across structuresData for historic BIM (HBIM)

  • Fast cataloguing
  • Comparative analysis across structures
  • Data for historic BIM (HBIM)

Does any of these cases fit?

Let's talk to assess feasibility and design a proof of concept.

Contact → Laboratory