基于目标检测算法的历史建筑清水砖墙劣化特征智能识别与评估诊断

Intelligent Identification and Diagnostic Assessment of Deterioration Features in Exposed Brick Walls of Historical Buildings Based on Object Detection Algorithms

通过深度学习和计算机视觉技术的融合应用,结合迁移学习优化神经网络架构,基于目标检测算法,实现对历史建筑清水砖墙劣化损伤特征的自动识别及精确评估。建立清水砖墙劣化特征样本数据库,选用PP-PicoDet算法进行模型训练与特征增量叠加训练,实现对损伤类型、位置和程度的快速且精确的自动识别。相较传统人工勘查模式,大幅提高了作业效率及结果准确性,为历史建筑保护修缮提供了新的智能化高效勘察工具和方法。

Automatic identification and accurate evaluation of deterioration features in exposed brick walls of historical buildings are achieved by integrating deep learning with computer vision technology,alongside optimizing the neural network architecture through transfer learning,based on object detection algorithms.A sample database of deterioration characteristics for exposed brick walls is established,and the PP-PicoDet algorithm is employed for model training and incremental feature enhancement.This approach enables rapid and precise automatic identification of damage types,locations,and severity.Compared to traditional manual inspection methods,this significantly enhances operational efficiency and result accuracy,offering a new intelligent and efficient tool and method for the protection and repair of historical buildings.