With the swift advancement of the social economy, traditional rural architectural heritage is gradually disappearing from people's sight. Rural architectural heritage carries the history and culture of human civilization development, and its digital protection has become increasingly important. To handle the issue of low detection accuracy in building inspection in the current digital technology for rural architectural heritage protection, an improved You Only Look Once version 8 (YOLOv8) algorithm is constructed, and a digital protection framework for rural architectural heritage based on PCA and improved YOLOv8 is constructed. Comparing the performance of improved YOLOv8 with other algorithms, it is found that the algorithm's frame rate per second and average accuracy are 69.4 frames/second and 98.69%, respectively, which are superior to the compared algorithms. Subsequently, a performance comparison analysis is conducted between the constructed digital protection framework for rural architectural heritage and other frameworks. The outcomes show that the recognition accuracy of the framework for four types of building components is 98.1%, 97.3%, 98.2%, and 97.8%, respectively, all higher than the comparison framework. The above results indicate that the improved YOLOv8 algorithm and digital protection framework for rural architectural heritage proposed in the study have good performance and can provide a theoretical basis for research in the field of architectural heritage protection.