The transformers revolutionized the field of deep learning, which is traditionally dominated by CNN. In contrast to normal photographs, hyperspectral imaging captures several narrow and consecutive spectral bands across the electromagnetic spectrum, providing 3D spatio-spectral data. CNNs are adept at extracting local information but are constrained to a limited receptive field. While transformers provide strong global representations, excluding finer details. The Swin Transformer architecture was proposed, with PCA applied beforehand for dimensionality reduction. It allows one to find self-attention within the windows and multi-head attention with the help of shifted windows to retain the attention between them, which is a drawback in the Vision Transformer (ViT). The model is integrated by layer normalization and MLP layers to enhance model stability. The developed model is evaluated on public datasets (IndianPines, Pavia University, Salinas), which obtained test accuracies of 98, 99, and 99%, as shown by the experimental findings using AdamW optimizers. It resulted in good generalization capabilities and capturing minute features with increased accuracy.