Crop classification is vital for modern agriculture, supporting applications such as early crop monitoring and efficient irrigation planning. However, distinguishing between different crop types remains a complex task for policymakers and analysts. With the increasing availability of satellite imagery that offers high temporal and spectral resolution, combined with advances in machine learning, large-scale automated crop monitoring and land-use assessment are now more achievable than ever.

The Prithvi-100M-multi-temporal-crop-classification model, developed by NASA and IBM, is a fine-tuned version of the Prithvi-100M earth observation foundation model. Trained on a multi-temporal crop classification dataset, this model enables automated identification and classification of crops from multispectral satellite imagery, significantly improving scalability, accuracy, and operational efficiency in agricultural monitoring.