Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{196170,
        author = {Ketan Kanjiya and Piyush Sonani and Upendrasinh Zala},
        title = {Automated Land Cover Classification from Satellite Images Using Deep Transformer Networks},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {7715-7724},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=196170},
        abstract = {Accurate land cover classification from satellite imagery is essential for environmental monitoring, urban planning, and sustainable resource management. Traditional machine learning and convolutional neural network approaches have achieved considerable success in remote sensing tasks; however, they often struggle to capture long-range spatial dependencies present in high-resolution satellite images. This study proposes a transformer based deep learning framework for automated land cover classification using semantic segmentation. The proposed approach utilizes the Mask2Former architecture with a Swin-Large backbone to perform pixel-level classification of multiple land cover categories. A patch-based training strategy is adopted to efficiently process large satellite images, while data augmentation techniques are applied to improve model generalization. To generate full resolution predictions, a sliding-window inference strategy is employed. The model is evaluated using standard semantic segmentation metrics, including Pixel Accuracy, Intersection over Union, Precision, Recall, and Dice coefficient. Experimental results show that the proposed framework achieves a mean Intersection over Union (mIoU) of 0.5701 and a pixel accuracy of 0.8716, indicating reliable segmentation performance across multiple land cover categories. These findings highlight the effectiveness of transformer-based architectures for large scale geospatial analysis and automated land cover mapping from satellite imagery.},
        keywords = {Deep Learning, Geospatial Analysis, Land Cover Classification, Remote Sensing, Satellite Imagery, Semantic Segmentation},
        month = {April},
        }