Satellite-based crop type identification plays a crucial role in enhancing food security and resource management. By accurately identifying crop types, it becomes possible to evaluate food production, predict shortages, and plan necessary interventions. Additionally, it enables efficient resource management, such as optimizing water and fertilizer usage, which promotes sustainability and minimizes waste. In this research, an end-to-end, deep learning (DL)-driven framework is proposed. The necessary steps for preprocessing Sentinel-2 images to generate training and testing datasets are first outlined. Next, the UNET architecture, utilizing a MobilNetv2 backbone, is adjusted and applied to segment the final image into 10 distinct classes: cotton, dates, grass, lucerne, maize, pecan, vacant, vineyard, vineyard-pecan (“Intercrop”), and background. The experiment resulted in an accuracy of 69%, with a macro F1 score of 70.8%