Modeling the functions and properties of DNA sequences is a challenging task in genomics, and it becomes even more complex when dealing with the human genome. The human genome consists of exons and introns, which have not yet been fully identified. This complexity arises from the fact that 98% of the human genome is composed of introns. Therefore, a highly accurate predictive model can provide significant advantages in advancing human genome research. In this study, we represented human exon and intron sequences as RGB images to facilitate their classification. We implemented a convolutional neural network (CNN) architecture to classify two constructed databases. Our model achieved high classification performance, with training and testing accuracy exceeding 92%.