seminar by Dr.Rozhgar Qarani
Title: Building Extraction in Occluded Areas Using Hybrid Deep Learning
Description:
-Focuses on building extraction from satellite images in occluded areas
-Addresses challenges like trees, shadows, and overlapping objects
-Uses a hybrid deep learning approach:
-U-Net / FCN → detect building regions
-Mask R-CNN → refine and separate buildings
-Applies a sequential process for improved results
-Achieves better accuracy and boundary precision
-Suitable for complex urban environments and real-world applications
