Building Structure Damage Detection Method Using Convolutional Neural Network

In modern building structures, damage detection is an important link to ensure safety and extend service life. At present, damage detection of building structures is inaccurate and inefficient, making it prone to errors. This paper aims to study a CNN-based building structure damage detection method. By collecting various types of building damage image data sets, using data enhancement technology to expand the sample size, and using the CNN model for training and optimization. Specific steps include building the CNN model architecture, training using the cross-entropy loss function, adjusting the learning rate through the Adam optimizer, and finally evaluating the model performance on the test set. The proposed method achieved the highest accuracy of 92.3% and recall rate of 90.5% in the damage detection task, significantly better than traditional methods. The convolutional neural network method proposed in this study has broad application prospects.