Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Architectural Decoration Image Recognition and Classification based on Convolutional Neural Network
Architectural decoration images are of various types and complex features. Therefore, fast and accurate recognition and classification of architectural decoration images has become a key challenge to improve design efficiency and project management level. This paper proposes an innovative solution - a CNN-based architectural decoration image recognition and classification method. The paper first collects and annotates various types of architectural decoration images to construct a dataset containing multiple architectural decoration styles and preprocesses these images. In the construction of CNN model, a deep network structure integrating convolutional layer, pooling layer, fully connected layer, etc. is designed. The CNN model is trained using the preprocessed architectural decoration image dataset, and the model parameters are adjusted through the back propagation algorithm to minimize the classification error. The results show that the average recognition accuracy of CNN in architectural decoration image recognition reaches 97.61%, which is significantly higher than the 91.35% of DeepLabv3+, and the processing speed is also faster. This paper provides a new technical means for the field of architectural decoration, which helps to promote technological progress and innovation in this field.
Architectural Decoration Image Recognition and Classification based on Convolutional Neural Network
Architectural decoration images are of various types and complex features. Therefore, fast and accurate recognition and classification of architectural decoration images has become a key challenge to improve design efficiency and project management level. This paper proposes an innovative solution - a CNN-based architectural decoration image recognition and classification method. The paper first collects and annotates various types of architectural decoration images to construct a dataset containing multiple architectural decoration styles and preprocesses these images. In the construction of CNN model, a deep network structure integrating convolutional layer, pooling layer, fully connected layer, etc. is designed. The CNN model is trained using the preprocessed architectural decoration image dataset, and the model parameters are adjusted through the back propagation algorithm to minimize the classification error. The results show that the average recognition accuracy of CNN in architectural decoration image recognition reaches 97.61%, which is significantly higher than the 91.35% of DeepLabv3+, and the processing speed is also faster. This paper provides a new technical means for the field of architectural decoration, which helps to promote technological progress and innovation in this field.
Architectural Decoration Image Recognition and Classification based on Convolutional Neural Network
Zhang, Wenhao (Autor:in) / Yang, Jianyang (Autor:in) / Yang, Mingjie (Autor:in)
24.01.2025
245779 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch