Hyperspectral image (HSI) classification has become a fundamental task with diverse applications in fields such as remote sensing, environmental monitoring, and precision agriculture. Traditional machine learning techniques and convolutional neural networks (CNNs) have been widely employed for HSI classification, but they often struggle to effectively capture the intricate spectral information inherent in these data. The sequential nature of spectral data is not fully exploited by conventional CNN architectures. In this paper, we are applying Transformer approach to HSI classification by leveraging the powerful capabilities of Transformer networks. Unlike traditional CNN models, Transformers excel at capturing sequential dependencies, making them well-suited for modeling the complex spectral information present in hyperspectral images. Our proposed method, Transformer-based Spectral-Spatial Feature Extraction, utilizes Transformers to learn local sequence information from neighboring bands, thereby enhancing the representation of spectral signatures. Furthermore, this approach incorporates cross-layer skip connections to retain important spectral features throughout the network, further improving the classification performance.