NB-NN-RF: an ensemble machine learning approach for classification of satellite images

• Published in: International journal of information technology (Singapore. Online) Vol. 17; no. 7; pp. 4093 - 4102
• Main Authors: Thakur, Reena, Borkar, Pradnya, Kothari, Varsha, Sharma, Gaurav, Singh, Vinay Kumar, Badhiye, Sagarkumar, Shelke, Sudhir
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 01.09.2025; Springer Nature B.V
• Subjects: Accuracy; Artificial Intelligence; Computer Imaging; Computer Science; Datasets; Image analysis; Image classification; Image Processing and Computer Vision; Land cover; Land use; Machine Learning; Neural networks; Original Research; Pattern Recognition and Graphics; Remote sensing; Satellite imagery; Scientific visualization; Software Engineering; Source programs; Vision; Visualization; NN; NB; RF; EuroSAT; Orange; ML; Image classification
• ISSN: 2511-2104; 2511-2112
• DOI: 10.1007/s41870-025-02619-5

Satellite image analysis requires computational proficiency and integration of data visualization strategies. However, traditional techniques are unable to classify satellite images due to a lack of interoperability issues. Thus, the given paper introduces an ensemble machine learning (ML) approach, utilizing the Orange tool to undertake image classification tasks specifically with the EuroSAT dataset. Orange is a free and open-source program used for pre-processing, visualization, and modeling of data. The classifiers used in the proposed approach are Naïve Bayes (NB), Neural Networks (NN), and Random Forest (RF). The EuroSAT dataset consists of satellite images of land cover types across Europe, with several classes ranging from natural landscapes to urban areas. The results show that the Orange tool, combined with the proposed approach (NB + NN + RF), can effectively classify EuroSAT images with high accuracy and performance, making it a valuable tool for remote sensing and land use applications. The proposed approach achieves the highest classification accuracy, thereby surpassing existing ML techniques.