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{188663,
        author = {Mr. Abhale B.A and Mr. Wadghule Y.M and Miss.Chaudhari Nisha and Miss.Pagar Shruti and Miss.Nikam Aarti and Miss.Darade Nikita},
        title = {Automated Dilated Convolutional Attention for Land Use Classification},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {12},
        number = {7},
        pages = {3276-3281},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=188663},
        abstract = {Accurate and timely Land Use and Land Cover (LULC) classification is a cornerstone of environmental management, urban planning, and sustainable resource allocation. Traditional methods, reliant on manual digitization or classical machine learning, are inefficient, slow, and struggle with the scale and complexity of modern remote sensing data. While Deep Learning (DL), particularly Convolutional Neural Networks (CNNs), has significantly advanced LULC mapping, a critical operational gap remains: data siloing. Most DL models are designed for a single modality (e.g., 3-channel RGB or 4-channel multispectral) and produce outputs tied to a specific dataset's legend. This survey paper explores the evolution of LULC classification, focusing on the challenge of multi-modal data ingestion and output harmonization. We review foundational architectures like U-Net and DeepLabV3+ and investigate the use of dilated convolutions and attention mechanisms for high-resolution semantic segmentation. The core focus is on a hybrid architecture that intelligently routes multi-modal inputs (3-channel vs. 4-channel) to specialized models, yet produces a single, unified, and harmonized classification map. This study concludes that such a "router-based" hybrid approach is essential for creating a truly robust, scalable, and user-friendly LULC system that can adapt to diverse, real-world data sources.},
        keywords = {Land Use Land Cover (LULC), Deep Learning, Semantic Segmentation, High-Resolution Imagery, MultiModal Data, Data Fusion, Data Harmonization, U-Net, DeepLabV3+, Dilated Convolutions, Attention Mechanism, Multispectral, RGB, Remote Sensing, Input Router.},
        month = {December},
        }