Hyperspectral image (HSI) classification is challenging due to the data’s vast spectral and spatial information. This research presents a new method of classification that utilizes the attention mechanism of dual convolutional neural networks (CNN). The integrated attention mechanism helps the model prioritize attributes, drop irrelevant information, and maximize feature extraction. This method improves HSI classification by concatenating spatial and spectral data into a fully connected layer using an adaptive approach, improving classification over state-of-the-art approaches. The Indian Pines and Pavia University datasets are used to assess performance using overall accuracy, average accuracy, and kappa value. The study also examines how parameter changes affect model performance and compares results to other approaches. The study shows how attention processes can improve model performance, thereby improving the field and driving future research toward more accurate and efficient HSI classification.