🧠 Current Mission: Building reliable, uncertainty-aware ML systems for spatial cognition and urban monitoring
🛰️ Research Focus: Sketch map analysis, remote sensing, multimodal machine learning, conformal prediction
🎓 Base: Institute for Geoinformatics (IFGI), WWU Münster, Germany
- 📄 Introducing the VISQAM Dataset: Toward Automated Map Interpretation @ GeoAI 2026 — benchmark of 1,200 annotated thematic maps, 4,594 QA pairs; fine-tuning lifted BERTScore-F1 from 0.43 → 0.72
- 📄 An approach for verifying terrestrial photographs via geosignatures — accepted at COSIT 2026
| 🗺️ Sketch Map Analysis | 🛰️ Remote Sensing | 📊 Uncertainty Quantification |
|---|---|---|
| VQA on thematic/geographic maps, geosignature misinformation detection | Sentinel-1 SAR + Sentinel-2 optical fusion for urban monitoring (Delhi) | Conformal prediction, sensor-dropout robustness in digital twins |
🗺️ GMDA-Feature
A new feature for the SketchMapia software, based on Aaron L. Gardony's research on bi-dimensional regression. Stack:
JavaScript
Urban heat analysis for Delhi using remote sensing data. Stack:
Python
🧩 SketchMapia-Microservices (internship, IFGI)
Feature development and engineering on a microservice-based sketch-map analysis platform — extended results tables, added CSV export, integrated collaborator features, merged via fork-based PR.
