Deep learning and IoT-driven protection and inheritance of architectural heritage in traditional villages

Environmental erosion, ageing and human activities are significant factors in speeding up the deterioration of traditional villages, which also serve as the protectors of cultural heritage. In conventional protection practices, protection is done by manual inspection and sparse records, which cannot provide real-time monitoring and predictive capability. Given this, an intelligent protection and inheritance strategy for architectural heritages in typical villages is proposed based on deep learning (DL) and the Internet of Things (IoT). The IoT sensing network gathers information on the environment, the structure and the image, generating 408 image records of the heritage buildings with environmental and structural data. Wavelet transform is employed to denoising the acquired signals, adaptive filter adapts parameters dynamically to enhance the quality of the acquired signal, and spatiotemporal feature alignment is used for synchronizing signals in space and time to ensure that the signals are represented in the same way. For fine-grained perception, DL architecture integrating Namib Beetle Optimized Vision Transform with bidirectional Long Short-Term Memory (NB-ViT-BiLSTM) with NB is used to optimize the model parameters and improve learning efficiency. While ViT uses the spatial characteristics of the images of buildings to recognize its components and deteriorations, BiLSTM utilizes the temporal connections in the collected information to research the trend of structural degradation and possible risk predictions. This process of implementing and testing the proposed model was done using Python. From experimental results, it was observed that the proposed framework was able to detect damages with 98.78% accuracy and architecture components.