Transformer-based land use and land cover classification with explainability using satellite imagery

Summary

The article explores how explanations provided by 'Explainable AI' (XAI) enhance the interpretability of various AI models used for land use and land cover (LULC) classification. It compares transformer-based models' performance on datasets such as EuroSAT, integrating interpretability tools like Captum to pinpoint influential image regions and ensure accuracy and transparency in predictions, demonstrating the effectiveness of XAI in practical applications. Tags: 'Artificial neural networks, Machine learning, Image processing, Vision transformer, Deep learning, Large language models, Fine-tuning, Neural network, Training, validation, and test data sets'