The classification of facies in seismic sections involves various methods; one of the most common is the use of pre-stack inversion attributes. A support vector machine model was trained using the inversion attribute and well log facies, followed by validation using a well process. In the next stage, a convolutional neural network was employed, where 2D convolution processing was used to calculate feature maps on the seismic section, and the regression process was carried out through neural networks. The regression output was the same attributes mentioned in the support vector machine classification method. Subsequently, by combining the convolutional neural network with a pre-trained support vector machine model, a new hybrid model for facies classification was created. The validation accuracy of both models in classification was estimated to be 93 percent.