Accurate and timely information on crop distribution is crucial for decision- making in agriculture and ensuring global food security. Crop mapping using remote sensing data has become an essential tool for agricultural monitoring and management. The process of crop mapping involves the acquisition of mul- tispectral data from satellites, pre-processing of the data and analysis to iden- tify different crop types based on their spectral signatures. This information is then combined with ground truth data to create accurate crop mappings that show the location and extent of different crops within an area. In recent years, Convolutional Neural Network (CNN) models have been used for crop mapping using Sentinel-2 data. However, CNN models may not be effective in capturing the spatial dependencies between features extracted from multispectral data. To address this issue, we propose a transformer model. The proposed trans- former model is compared with the CNN model to demonstrate its effectiveness and accuracy for crop mapping. This study demonstrates the potential of the Transformer model in capturing the spatial dependencies between features and efficiently processing long sequences of data, contributing to improved agricul- tural practices, resource management and food security.