Machine Learning for ICT infrastructure detection

  • * Register (or log in) to the Neural Network to add this session to your agenda or watch the replay

  • Date
    15 September 2026
    Timeframe
    14:00 - 15:00 CEST, Geneva
    Duration
    1 hour

      This session introduces how ITU/BDT/DNE/FNS is applying machine learning (computer vision) to improve the availability and quality of ICT infrastructure data by detecting telecommunication towers from satellite imagery. Participants will see the end-to-end workflow (data labeling → model training → inference), learn how the work is being advanced with partners (including KAUST), and discuss how the resulting data and capacity-building approach can strengthen ITU/BDT connectivity programmes and country support. 
       
      Session Objectives

      By the end of this session, participants will be able to:

      • A clear understanding of the problem in getting global connectivity, addressing: gaps in reliable tower/infrastructure data for network planning and gap analysis.
      • A practical view of the tower-detection pipeline: image collection and annotation, ML model training, and inference to extract tower locations.
      • Awareness of the supporting tools developed for demonstration and training (labeling tool and tower detector app) and how they can be used in capacity-building contexts.
      • An overview of how partnerships (e.g., KAUST) contribute to model improvement, scaling, and knowledge transfer.
      • A shared discussion on how improved infrastructure data and AI-enabled workflows can feed into ITU/BDT programmes (planning, measurement, and implementation support) and where to pilot next. 

      Are you sure you want to remove this speaker?