The challenge
Every year, undetected road degradation causes thousands of accidents across the African continent. Manual inspection methods are costly, slow and impossible to deploy beyond a few priority routes. Authorities are navigating blind: no tool allows them to know, in real time, where a road is degrading, how fast, and which sections to prioritize.
The vision: GEOWATCH
GEOWATCH is an intelligent geospatial monitoring ecosystem for African infrastructure. It is not a single piece of software, but a modular platform where each branch covers a specific monitoring domain.
Road Intelligence is the first project of the GEOWATCH program. Its mission: cover the entire road monitoring lifecycle, from degradation detection to severity classification, from evolution prediction to decision support for infrastructure managers. The ambition is to make every kilometer of African road observable from space.
Project genealogy
TerraPulse Vision (2023) The founding research topic. Born from a road accident I suffered in 2020, caused by a pothole hidden under mud. In 2023, I began formalizing this experience into a research topic: prove that a machine learning model trained on satellite imagery can automatically detect road degradation in the African context.
GeoSmart Vision (April-May 2025) The validation point. In 2025, my university's FabLab was looking for a topic on mapping and territorial monitoring to participate in MASS 2025. I proposed TerraPulse Vision. The university validated it and presented it under the name GeoSmart Vision, with me remaining as project lead and owner. The 2nd Prize at the African Market for Space Solutions validated the research hypotheses and confirmed the topic's viability.
Road Intelligence The realization. This is the transformation of the TerraPulse Vision research topic into a fully operational project, and the first project of the GEOWATCH program. Road Intelligence takes TerraPulse Vision's scientific core and deploys it as an actionable solution: real-time detection, decision support, interface for infrastructure managers.
Technical architecture
The ecosystem is built around three layers:
Acquisition Multispectral satellite imagery (Sentinel-2), OpenStreetMap data, and ground truth from sensors and direct observations.
Processing Machine learning pipeline integrating supervised classification and image segmentation for detection and geolocation of degradation. Models are trained exclusively on African ground truth data.
Output Real-time web cartographic platform designed for decision-makers: degradation zone visualization, impact assessment, intervention prioritization.
The deliverable
A functional prototype demonstrating automatic road degradation detection at regional scale. Complete technical documentation: functional specifications, technical specifications, quality plan. Cartographic visualization interface. Architecture extensible to other infrastructure types.
My contribution
Complete architecture design of the GEOWATCH ecosystem. Original research topic formulation (TerraPulse Vision). Full authorship of technical specifications. Machine learning pipeline development. Team leadership and deliverable coordination for MASS 2025.
Recognition and impact
Two official invitations followed: the GMES & Africa Phase II Continental Forum in Cairo (December 2025), signed by H.E. Dr Tidiane Ouattara, Chairman of the African Space Agency Council, and the NewSpace Africa Conference 2026 in Libreville, with a joint invitation from AfSA and AGEOS.
Technical resources
Technology badges are displayed below.
- Python
- TensorFlow
- scikit-learn
- Sentinel-2
- PostGIS
- PostgreSQL
- QGIS
- OpenStreetMap
- Next.js
- TypeScript