This paper proposes a deep convolutional neural network (CNN) for hyperspectral image (HSI) classification. The proposed CNN classifier consists of five layers: input, convolutional, max pooling, full connection, and output. The architecture is designed to directly classify HSI data in the spectral domain. Experimental results on three HSI data sets (Indian Pines, Salinas, and University of Pavia) show that the proposed method outperforms traditional methods such as support vector machines (SVM) and conventional deep learning approaches. The CNN achieves higher classification accuracy, especially for small training samples. The proposed CNN is effective for HSI classification, with a simple architecture that includes one convolutional layer and one fully connected layer. The method is trained using back propagation and gradient descent, with parameters initialized randomly. The CNN is evaluated on three data sets, demonstrating its effectiveness in classifying HSI data. The results show that the proposed CNN achieves better performance than SVM and other neural networks, with higher accuracy and faster training and testing times. The paper also discusses future work, including the use of Siamese networks and techniques like dropout to improve performance. The study highlights the potential of deep learning, particularly CNNs, for HSI classification.This paper proposes a deep convolutional neural network (CNN) for hyperspectral image (HSI) classification. The proposed CNN classifier consists of five layers: input, convolutional, max pooling, full connection, and output. The architecture is designed to directly classify HSI data in the spectral domain. Experimental results on three HSI data sets (Indian Pines, Salinas, and University of Pavia) show that the proposed method outperforms traditional methods such as support vector machines (SVM) and conventional deep learning approaches. The CNN achieves higher classification accuracy, especially for small training samples. The proposed CNN is effective for HSI classification, with a simple architecture that includes one convolutional layer and one fully connected layer. The method is trained using back propagation and gradient descent, with parameters initialized randomly. The CNN is evaluated on three data sets, demonstrating its effectiveness in classifying HSI data. The results show that the proposed CNN achieves better performance than SVM and other neural networks, with higher accuracy and faster training and testing times. The paper also discusses future work, including the use of Siamese networks and techniques like dropout to improve performance. The study highlights the potential of deep learning, particularly CNNs, for HSI classification.