Overview

A satellite product is only as credible as its validation. Trained as a geomatic engineer, I design and carry out the GNSS, Total Station and UAV surveys that calibrate and test Earth observation models, and I report accuracy in terms decision-makers understand.

What you get

  • Validation campaign design: what to measure, where and how many samples
  • GNSS and Total Station ground-control networks
  • UAV photogrammetry, orthomosaics and labels for model training
  • Participatory plot-level ground-truth protocols
  • Accuracy assessments comparing EO outputs with field measurements

Who this helps

  • EO and AI projects

    That need independent evidence that their maps are right.

  • Mining & infrastructure

    Survey control for validating satellite- and drone-derived models.

  • Transport agencies

    Metric-accurate road assessment from low-cost sensors.

  • Research teams

    Designing validation that will stand up to peer review.

How I approach it

  1. 01

    Design validation before collecting data

    Choose the error metric and a sampling design stratified across conditions, so validation is not biased toward easy areas.

  2. 02

    Establish survey control

    GNSS receivers and Total Stations set ground-control points that tie everything to a known reference.

  3. 03

    Capture at the right scale

    UAV flights provide high-resolution orthomosaics and labels where satellite resolution is not enough.

  4. 04

    Involve local knowledge

    Participatory plot-level ground truth brings in what only people on the land know.

  5. 05

    Compare like with like

    Co-register field and EO data carefully; the TRECK road study compared outputs with 50 surveyed measurements along 50 km.

  6. 06

    Report accuracy honestly

    State errors, coverage and known gaps, so users know where a product can be relied on.

Evidence from my work

  • 50 / 50 kmSurveyed ground-truth measurements / road length (TRECK)
  • 333.9 km²Monitored with participatory ground truth
  • First ClassBSc Geomatic Engineering, KNUST
  • 88%Best Final Year Project, sensor fusion & computer vision

Field experience includes establishing ground-control points with Total Stations, GNSS and UAVs at Asanko Gold Ghana (2023).

Selected work

Sensor FusionIEEE T-ITS

Multi-Modal Road Assessment (TRECK)

Camera / IMU / GPS fusion for metric-accurate road assessment, validated against 50 surveyed ground-truth measurements along 50 km of road.

View case study
UAVObject Detection

Cocoa Plant Detection & Counting

YOLOv8 with multi-temporal analysis and ensemble uncertainty scoring to detect and count cocoa plants across smallholder and commercial layouts.

View case study
SAR–Optical FusionYOLOv8

Atewa Illegal Mining Detector

Dual-stream U-Net + YOLOv8 fusing Sentinel-1 and Sentinel-2, trained on UAV-derived labels. Recall 82%, F1 0.79 on held-out tiles.

View case study

What sets my work apart

  • 1

    Engineer and data scientist

    I understand both the survey instrument and the model it validates.

  • 2

    Validation designed in

    Sampling and metrics are planned at the start of a project, not after the map is finished.

  • 3

    Metric accuracy

    Results expressed in real-world units, such as metres, centimetres and counts, rather than abstract scores alone.

Tools & standards

  • GNSS
  • Total Station
  • UAV photogrammetry
  • Ground-control networks
  • Sensor fusion (camera / IMU / GPS)
  • QGIS
  • ArcGIS Pro

Common questions

Can you validate a product someone else built?

Yes. Independent validation is often the most credible kind; I design the sampling and comparison around your product and its intended use.

Do you need to be on site?

For GNSS, Total Station and UAV work, yes, or a local team can follow a protocol I design. Analysis and reporting can be remote.

What does an accuracy report contain?

The sampling design, error metrics with confidence, maps of where errors occur, and recommendations on where the product should not be used.