{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,11,9]],"date-time":"2025-11-09T05:55:50Z","timestamp":1762667750424,"version":"build-2065373602"},"reference-count":0,"publisher":"Zarqa University","issue":"6","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IAJIT"],"published-print":{"date-parts":[[2025]]},"abstract":"<jats:p>The rapid changes in Deep Learning (DL) have made better performing models for Remote Sensing Images (RSIs), particularly for semantic segmentation and multi-class classification. This study looks at how two common DL structures, U-Net and DeepLabV3+, perform on segmentation tasks, while 201-layer Densely Connected Convolutional Network (DenseNet201-CNN) is compared to Visual Geometry Group 16-layer network (VGG16) for classification tasks. The dataset for segmentation has aerial images of Dubai labeled with pixel-level segmentation across six classes: Building, land, road, vegetation, water, and unlabeled. The classification data set called Remote Sensing Image-Collection of Benchmark 256 (RSI-CB256) has Remote Sensing (RS) pictures sorted into four groups: Cloudy, desert, green_area, and water. DeepLabV3+ demonstrated better training performance and convergence behavior compared to U-Net, exhibiting more stable learning and efficient boundary detection during segmentation. While both models performed competitively, DeepLabV3+ consistently showed stronger generalization capability, making it more effective in delineating complex land cover boundaries. In contrast, U-Net displayed sensitivity to hyperparameters and greater variation in performance, indicating the need for further tuning and regularization. U-Net was good initially but had varied performance and was sensitive to hyperparameters suggesting the need of better regularization techniques. In terms of the classification, DenseNet201-CNN did better than VGG16 in precision, recall, and F1-score for all categories. Notable performance gains were observed in \u201ccloudy\u201d and \u201cdesert\u201d classes where DenseNet201-CNN model demonstrated significantly fewer misclassifications. Overall, DenseNet201-CNN outperformed VGG16 in terms of total classification accuracy. These results establish DenseNet201-CNN as a superior choice in RSI classification tasks in this study<\/jats:p>","DOI":"10.34028\/iajit\/22\/6\/7","type":"journal-article","created":{"date-parts":[[2025,11,5]],"date-time":"2025-11-05T07:58:50Z","timestamp":1762329530000},"source":"Crossref","is-referenced-by-count":0,"title":["Evaluation of Deep Learning Models for Remote Sensing Segmentation and Classification"],"prefix":"10.34028","volume":"22","author":[{"given":"SaiVenkataLakshmi","family":"Ananth","sequence":"first","affiliation":[]},{"given":"Suryakanth","family":"Gangashetty","sequence":"additional","affiliation":[]}],"member":"19944","published-online":{"date-parts":[[2025]]},"container-title":["The International Arab Journal of Information Technology"],"original-title":[],"language":"en","deposited":{"date-parts":[[2025,11,9]],"date-time":"2025-11-09T05:51:20Z","timestamp":1762667480000},"score":1,"resource":{"primary":{"URL":"https:\/\/iajit.org\/upload\/files\/Evaluation-of-Deep-Learning-Models-for-Remote-Sensing-Segmentation-and-Classification.pdf"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025]]},"references-count":0,"journal-issue":{"issue":"6","published-online":{"date-parts":[[2025]]},"published-print":{"date-parts":[[2025]]}},"URL":"https:\/\/doi.org\/10.34028\/iajit\/22\/6\/7","archive":["Internet Archive"],"relation":{},"ISSN":["2309-4524","1683-3198"],"issn-type":[{"type":"electronic","value":"2309-4524"},{"type":"print","value":"1683-3198"}],"subject":[],"published":{"date-parts":[[2025]]}}}