Semantic Segmentation using Deep Learning - Meenu Vijarania - Bog

This research addresses multifaceted environmental challenges linked to urbanization by introducing a transferable UNET model designed for precise greenery detection in high-resolution satellite imagery. The dataset, consisting of 72 images captured over Dubai by MBRSC satellites and annotated by the Roia Foundation, serves as the foundation of our study. While rooted in Dubai, the model's adaptability extends beyond regional confines, offering a versatile solution to a spectrum of global urban environmental challenges. In addition to accurate greenery detection, the UNET model provides a holistic approach to urban land cover classification.