Wyniki wyszukiwania - BazTech

Ograniczanie wyników

2 Advances in Geodesy and Geoinformation

2 2025

Znaleziono wyników: 2

Liczba wyników na stronie

Wyniki wyszukiwania

Sortuj według:

Ogranicz wyniki do:

Integration of LIME with segmentation techniques for SVM classification in Sentinel-2 imagery

Yilmaz Elif Ozlem, Uzun Yakup Kaan, Berkan Emre, Kavzoglu Taskin

Advances in Geodesy and Geoinformation

2025

Vol. 74, no. 1

art. no. e61, 2025

Recent advancements in remote sensing technology have facilitated the acquisition of images with higher spatial resolution. In response to this rapid technological evolution, the paradigm of OBIA has emerged as a key approach. An essential component of OBIA is image segmentation, where the careful selection of an appropriate segmentation algorithm and its parameters significantly influences the quality of the segmentation output. This study aims to conduct LULC analysis on Sentinel-2 imagery and compare the accuracy of the SVM classifier across different segmentation methods produced by MRS, SLIC, Mean Shift, and Quick Shift algorithms. The selected study area is located in the Marmara region of Turkey and characterized by seven major LULC classes. The segmentation was conducted though four algorithms, with 60 segment features being extracted for each output, considering spectral, textural, and geometric attributes separately. Following the classification process with SVM, overall accuracies of 96.14% for the MRS, 91.00% for the SLIC, 89.95% for the Mean Shift and 87.95% for the Quick Shift approach were estimated. These results underscore the superior performance of the MRS algorithm with significant level of improvement. This high level of accuracy holds significant potential for delivering more dependable and precise outcomes in planning and decision-making processes. Moreover, integrating XAI, specifically the LIME algorithm, enhances the transparency and comprehensibility of classification analysis within the OBIA framework. Features associated with the NIR and SWIR bands were found to have predominantly positive effects. This integration contributes to improved transparency, enabling more informed and reliable decision-making processes.

Global-scale explainable AI assessment for OBIA-based classification using Deep Learning and Machine Learning methods

Kavzoglu Taskin, Uzun Yakup Kaan, Berkan Emre, Yilmaz Elif Ozlem

Advances in Geodesy and Geoinformation

2025

Vol. 74, no. 1

art. no. e62, 2025

Over the past decade, object-based image analysis (OBIA) has gained prominence as a widely adopted method for generating land use/land cover (LULC) maps. This study aims to evaluate the performance of various classification algorithms within the OBIA framework using SPOT-6 satellite imagery. The research methodology involved segmenting the images with the multi-resolution segmentation (MRS) algorithm, followed by the application of convolutional neural networks (CNN), random forest (RF), and support vector machine (SVM) algorithms for classification. The study was conducted in the Perpignan province, located in the Pyrénées-Orientales region of France. After the segmentation stage, CNN, RF, and SVM classifiers were employed to classify the image segments based on both spectral and spatial attributes. The accuracy of the resulting thematic maps was assessed using standard metrics, including overall accuracy (OA), the Kappa coefficient (KC), and the F��score (FS). Of the three classifiers, CNN achieved the highest overall accuracy at 91.28%, outperforming SVM, which attained an OA of 90.50%, and RF, which recorded an OA of 87.28%. Additionally, this study explored the integration of explainable artificial intelligence (AI) techniques, specifically the Shapley Additive Explanations (SHAP) algorithm, to enhance the interpretability of the machine learning models. This approach fosters greater trust, accountability, and acceptance in decision-making processes. By leveraging SHAP values, the study provides deeper insights into the decision-making processes of the CNN, SVM, and RF classifiers, ultimately enhancing the transparency and comprehensibility of these models.