Cross-Verification of Revenue Using Remote Sensing Data
Remote-sensing and GIS decision support for identifying construction activity and supporting property-tax verification in Peshawar, Pakistan.
September 2021 – June 2022
This project explored how remote sensing could support verification of property-tax records in the Peshawar region of Pakistan. Working with the Technology for Pakistan Initiative, I helped train a machine-learning model to classify constructed and unconstructed land using satellite imagery from multiple years.
I prepared and preprocessed imagery for model training and helped integrate the resulting detections into a web-based GIS system for temporal and spatial comparison of construction patterns across union councils. The platform also supported plot-level analysis of constructed properties to help identify potential gaps in tax records.
The resulting pipeline achieved 92% accuracy in identifying constructed areas. The broader research contributed to the International Growth Centre report Cross Verification of Revenue Using Remote Sensing Data.
Areas: Computer Vision · Machine Learning · Remote Sensing · Human-Computer Interaction · GIS · Web Development
Code: GitHub — TaxData
Technical Report: Cross Verification of Revenue Using Remote Sensing Data — International Growth Centre
Additional Details: LinkedIn Projects