Mitigating class imbalance in deep learning-based multi-class structural damage recognition using an informatics-oriented data augmentation framework

Key Points

• The approach significantly enhances detection accuracy for underrepresented classes in structural damage recognition.
• Data augmentation techniques were crucial in addressing the class imbalance during the analysis, leading to better model performance.
• Using deep learning, the framework improved multi-class recognition of structural damage through enhanced data handling and variability.
• This method highlights the need for advanced data strategies to effectively manage class imbalance in real-world applications.