This paper explores the utilization of convolutional neural networks (CNNs) in classifying petroglyphs from the Maule Region - Chile. It emphasizes the importance of deep learning approaches for cultural heritage analysis. The methodology encompasses data preparation, network architecture setup, and parameter tuning, utilizing tools tailored for scientific analysis and machine learning. The study analyzes the implemented code and execution to derive meaningful results and performance metrics. The discussion highlights the potential of CNNs in enhancing interpretative precision and addressing challenges in petroglyph classification. Despite achieving promising accuracy, further investigation into model refinement and mitigation of biases is warranted for informed decision-making in classification tasks. The analysis of the petroglyphs of the Maule region not only sheds light on the cultural heritage of the communities, but also raises questions about cultural appropriation and conservation and preservation strategies to enhance it. The conclusions showed the need to improve the model to avoid over-fitting and to obtain better results in the classification of petroglyphs.