Revolutionizing Heritage Site Information Retrieval: A Deep Learning Approach Utilizing CNN and SVM for Effective Classification of Cultural Heritage Sites

Categorizing cultural heritage photographs into many categories, such as historical relevance, architectural features, and preservation status, is necessary for retrieving information from heritage sites. This research investigates categorising photographs from historical sites using deep learning methods like Convolutional Neural Networks (CNN) and Support Vector Machines (SVM). For classification, we created a system that used a CNN model with three convolutional layers, four max-pooling layers, a fully connected layer, and an SVM model. Our model classified the 8645 photos in the dataset utilized in this research into six separate categories, with an overall accuracy of 0. 89091. To create a successful historic site information retrieval system, comprehensive data collection and processing, feature extraction, algorithm creation, and assessment metrics are essential. We review the methods and strategies used in each process and the difficulties and constraints faced during the system's development and testing. To obtain high classification accuracy, thorough data processing and model optimization is crucial. Our results show the potential of deep learning techniques for historic site information retrieval. Future research might increase the dataset's size further to verify the efficacy of the CNN and SVM models to include additional categories and pictures. It may also be investigated to optimize the CNN and SVM models further to increase classification accuracy. Overall, our work offers insights into the significance of meticulous data processing and model optimization in creating an efficient system and adds to developing a heritage site information retrieval system utilizing CNN and SVM algorithms.