Tourism Image Retrieval Method Based on Deep Residual Shrinkage Network

In response to the problem of high image similarity in traditional tourism image retrieval, a tourism image retrieval method based on attention mechanism improved deep residual shrinkage network was proposed. The study first utilized deactivation mechanisms and activation functions to improve the deep residual shrinkage network, and applied it to feature extraction in tourism images. Then, based on the improvement of the network, attention mechanism was introduced to construct a retrieval model. The results showed that the average accuracy and consistency of the retrieval model in image dimensionality reduction were 95.18% and 89.55%, respectively. At the same time, the training loss value and image retrieval accuracy in both datasets were 0.21, 0.16, and 94.52% and 90.88%, respectively. And the precision and recall of the image were 92.61% and 96.48%, respectively, which were better than the comparison method. This indicated that the retrieval model could effectively improve the feature representation and retrieval performance of tourism images, while enhancing its robustness to noise and interference. The research aimed to provide a new and effective solution for tourism image retrieval.