Moudden et al. (2026) Deep Learning and Sentinel-2 Imagery for Crop Type Segmentation

Identification

• Journal: Lecture notes in networks and systems
• Year: 2026
• Date: 2026-01-01
• Authors: Tarik El Moudden, Mohamed Amnai, Ali Choukri, Prof. Youssef fakhri, Noreddine Gherabi
• DOI: 10.1007/978-3-032-12869-0_11

Research Groups

• Computer Science Research Laboratory, Ibn Tofail University, Kenitra, Morocco
• National School of Applied Sciences, Sultan Moulay Slimane University, Khouribga, Morocco

Short Summary

This study proposes an end-to-end deep learning framework utilizing a UNET architecture with a MobilNetv2 backbone and Sentinel-2 imagery for crop type segmentation, achieving 69% accuracy and a 70.8% macro F1 score across 10 distinct classes.

Objective

• To develop and apply an end-to-end deep learning framework for accurate crop type segmentation using Sentinel-2 satellite imagery to enhance food security and resource management.

Study Configuration

• Spatial Scale: Regional agricultural area (implied by crop type segmentation).
• Temporal Scale: Not explicitly defined, but implied multi-temporal for crop growth cycles using Sentinel-2 imagery.

Methodology and Data

• Models used: UNET architecture with a MobilNetv2 backbone (Deep Learning).
• Data sources: Sentinel-2 satellite imagery.

Main Results

• The proposed deep learning framework successfully segmented crop types into 10 distinct classes: cotton, dates, grass, lucerne, maize, pecan, vacant, vineyard, vineyard-pecan (“Intercrop”), and background.
• The model achieved an overall accuracy of 69%.
• A macro F1 score of 70.8% was obtained.

Contributions

• Proposal of an end-to-end deep learning framework specifically tailored for crop type segmentation using Sentinel-2 imagery.
• Application and evaluation of a UNET architecture with a MobilNetv2 backbone for this specific task.
• Demonstration of the potential of deep learning and Sentinel-2 data for agricultural monitoring and resource management.

Funding

• Not explicitly mentioned in the provided text.

Citation

@article{Moudden2026Deep,
  author = {Moudden, Tarik El and Amnai, Mohamed and Choukri, Ali and fakhri, Prof. Youssef and Gherabi, Noreddine},
  title = {Deep Learning and Sentinel-2 Imagery for Crop Type Segmentation},
  journal = {Lecture notes in networks and systems},
  year = {2026},
  doi = {10.1007/978-3-032-12869-0_11},
  url = {https://doi.org/10.1007/978-3-032-12869-0_11}
}