This study presents a land-cover classification (LAC) system using deep learning with high-resolution remote-sensing imagery. The research focuses on urban land use classification, where traditional remote-sensing methods face challenges due to the complexity and dynamic nature of urban land areas. To address these challenges, the study employs transfer learning with Inception-v3 and DenseNet121 architectures, leveraging pre-trained features from large datasets to enhance model generalization and performance. The fine-tuning process allows the model to adapt to the intricacies of land-cover classification, improving accuracy and reliability. The proposed system is evaluated on the UC-Merced_LandUse dataset, achieving high performance metrics, including 92% accuracy, 93% precision, 92% recall, and a 92% F1-score. The study also incorporates heatmap analysis to visualize the decision-making process of the models, providing insights into the classification mechanism. The results demonstrate the effectiveness of the proposed approach, highlighting the potential of deep learning in improving land-cover classification for urban planning, agricultural zoning, and environmental monitoring. The study compares the performance of Inception-v3, DenseNet121, and ResNet-50 models, finding that Inception-v3 outperforms the others in terms of accuracy and F1-score. The research also discusses the advantages of transfer learning and fine-tuning in optimizing model accuracy for complex land-cover classification tasks. The findings contribute to the evolution of remote-sensing methodologies and emphasize the importance of incorporating advanced techniques such as fine-tuning and specific network architectures in improving land-cover classification systems. The study concludes that the proposed LAC system is a robust and efficient solution for accurate and automated land-area classification.This study presents a land-cover classification (LAC) system using deep learning with high-resolution remote-sensing imagery. The research focuses on urban land use classification, where traditional remote-sensing methods face challenges due to the complexity and dynamic nature of urban land areas. To address these challenges, the study employs transfer learning with Inception-v3 and DenseNet121 architectures, leveraging pre-trained features from large datasets to enhance model generalization and performance. The fine-tuning process allows the model to adapt to the intricacies of land-cover classification, improving accuracy and reliability. The proposed system is evaluated on the UC-Merced_LandUse dataset, achieving high performance metrics, including 92% accuracy, 93% precision, 92% recall, and a 92% F1-score. The study also incorporates heatmap analysis to visualize the decision-making process of the models, providing insights into the classification mechanism. The results demonstrate the effectiveness of the proposed approach, highlighting the potential of deep learning in improving land-cover classification for urban planning, agricultural zoning, and environmental monitoring. The study compares the performance of Inception-v3, DenseNet121, and ResNet-50 models, finding that Inception-v3 outperforms the others in terms of accuracy and F1-score. The research also discusses the advantages of transfer learning and fine-tuning in optimizing model accuracy for complex land-cover classification tasks. The findings contribute to the evolution of remote-sensing methodologies and emphasize the importance of incorporating advanced techniques such as fine-tuning and specific network architectures in improving land-cover classification systems. The study concludes that the proposed LAC system is a robust and efficient solution for accurate and automated land-area classification.