Aisun Zarghami

I've reconstructed 2D surface and 3D height changes in Casa Grande, Arizona, using historical aerial imagery from 1961, 1978, and 1990.

For my M.Sc. thesis at the University of Tehran, I built a photogrammetric workflow in Agisoft Metashape for image alignment, ground-control-based georeferencing, orthophoto generation, and DSM production.

I developed Python geospatial workflows with PyTorch, GeoPandas, Rasterio, and scikit-learn, evaluating 10 segmentation architectures for building extraction. My selected Double Attention U-Net++ with ResNet50 reached F1 scores of 0.751 and 0.834 across historical imagery years.

I also build computer-vision projects spanning facial emotion recognition, genetic-algorithm circle detection, image-processing GUIs, YOLO object detection, and pollution prediction. I'm interested in applying photogrammetry, GeoAI, and spatial analysis to real-world geospatial problems.