Integrated Master, National Technical University of Athens (NTUA)
2020 – Current
School of Rural, Surveying and Geoinformatics Engineering
Using the mesogeos Dataset and training Deep Learning Models, such as UNet and Vision Transformers, a prediction of the final burned area is made for Wild Fire events. Dataset consists of 28 Remote Sensing variables, and Explainable AI is used to collect data for understanding Wild Fires.
Internship, National Observatory of Athens - Beyond Earth Observation Center
June 2024 – Sept 2024
Wild Fire Grading and Delineation from Sentinel 2 Data using QGIS and mostly Python
Worked at 2 Copernicus Emergency Management Service (EMS) projects
Developed Python Tools for Wild Fire mapping and statistics extraction such as Land Use, Land Cover, NATURA2000 etc
Developed Python tool for fully automated Wild Fire mapping to be used on EMS
Developed Machine Learning Dataset for Fire Risk Forecas
Burned Area mapping from Sentinel 2 Images
Developed in Python with Frameworks Xarray, GeoPandas, odc, Rasterio, Shapely etc.. Given an area (AOI box) with 4 coordinates and the start date of a wild fire it makes fully automatic mapping from Sentinel 2 data. Detailed information is available GitHub
Burned Area Statictics Calculation
Developed in Python, given a Shapefile polygon of burned area and Shapefiles such as NATURA2000, Corine Land Cover, etc., calculates percentage each shapefile is within the polygon of the fire. Detailed information is available GitHub
Horizonal Network Geodesy
Python script for solving 2D geodesy networks with complex geometry given only coords of the points in the network. Script solves the network and produces a pdf file with the report of the solving parameters and plots the shape of the given network – useful for finding problematic points within the network, info given in the pdf report, and remove them. Or finding problematic observations
Elevation Network Geodesy
Python script for solving 1D geodesy elevation networks given the ortho heights of the network. The script solves the network, print the corrected Heights, pdf report of the network parameters – useful for problematic height detection within the network
Mother tongue : Greek (Ελληνικά)
Foreign Languages : English, C2 Level (Proficiency)
Languages: Python, C++
Frameworks: Xarray, GeoPandas, Pandas, NumPy, Matplotlib, Pytorch
Programms: QGIS, ArcGIS, AutoCAD(2D), Git
Machine Learning in Geoinfomatics, basic theory of deep learing models and a semester topic of training a UNet to detect a specific object, such as detecting active fires from thermal camera data
Remote Sesning ΙI, theory and application of pixel based classification in Sentinel 2 Images and training classifiers with deep learing models with remote sensing data
Remote Sensing Ι, fundamendals of remote sensing and prossecing of data in QGIS, such as calculating indexs like NDVI, and learning to donwload LANDSAT data
Microwave Remote Sensing, theory of InSAR Images and applications like creating a Digital Terrain Model (DEM) from phase difference and detecting micromovements after earthquakes, Application in Sentiel 1 Images with SNAP software