图像检索模型训练方法、图像检索方法、设备和介质
By using an adaptive ensemble learning module and relation consistency loss, the overfitting problem in deep hashing methods is solved, enabling diversified representation and semantic aggregation of image features, thereby improving the accuracy and efficiency of image retrieval.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- WEIFANG UNIVERSITY
- Filing Date
- 2026-02-10
- Publication Date
- 2026-07-17
AI Technical Summary
Existing deep hashing methods suffer from overfitting in large-scale image retrieval, leading to a reliance on rigid patterns in the training data and making it difficult to effectively capture intra-class variations and generalize to unseen categories.
An adaptive ensemble learning module is employed to extract diverse features from images through multiple encoders and decoders. By combining reconstruction loss and relation consistency loss, dominant features are identified and the similarity relationships between images are strengthened, generating binary codes with semantic consistency.
It improves the discrimination and generalization capabilities of image retrieval models, enhancing retrieval accuracy and efficiency under large-scale image data.
Smart Images
Figure CN121706857B_ABST