As the service life of buildings increases, the problem of exterior wall damage has gradually become an important factor affecting the safety and service life of buildings. Previous methods of detecting damage to exterior walls of houses mostly relied on manual inspections, resulting in low detection efficiency and poor accuracy, which made it difficult to meet the needs of modern building maintenance. To this end, this study introduces a house building exterior wall damage detection and maintenance decision system based on the CNN algorithm. First, this paper uses a convolutional neural network (CNN) to extract features from exterior wall images and identify damaged areas. Then, this paper combines image processing technology and data enhancement methods to improve the robustness and accuracy of the model. Finally, this paper trains a multi-layer perceptron (MLP) model to classify damage types and automatically push reasonable maintenance plans based on the degree of damage. The experimental results show that the accuracy of the CNN model in damage detection reached 94.8%. In terms of maintenance decision push, the decision accuracy of the CNN model for mild and moderate damage is 92.5% and 85.3%, respectively. In the treatment of severe damage, although CNN performs better than traditional methods, it still needs further optimization to improve the ability to handle complex damage. In the above data conclusions, the research system can significantly improve the performance of external wall damage detection, providing an intelligent solution for building maintenance.