Optimizing the Classification and Identification Process of Archaeological Relics by Image Recognition Algorithm

With the deepening of archaeological excavations, the number of cultural relics has surged, and the traditional classification and identification methods are inefficient and subjective because they rely on expert experience and intuition. In this paper, a comprehensive image recognition framework based on deep learning is proposed. The core algorithm includes the combination of convolutional neural network (CNN) and transfer learning strategy, and a fine-grained classification algorithm optimized for specific cultural relics. By using high-resolution camera and professional scanning equipment to collect images and preprocessing them to enhance the accuracy and robustness of the model, this study verifies the effectiveness of the framework in the actual archaeological data set. The experimental results show that compared with the baseline model (ResNet-50), the model with attention mechanism has significantly improved in accuracy, precision, recall and F1 score, with the accuracy increasing from 88.0% to 92.0%, precision from 87.5% to 91.2%, recall from 88.5% to 92.4%, and F1 score from 88.0% to 90%. These results prove the potential of the proposed model in improving the efficiency and accuracy of archaeological cultural relics classification and identification, providing strong technical support for archaeological research, and providing useful reference for the application of image recognition technology in other related fields.