特征聚集 Hash 网络

In the era of information explosion, image retrieval technology plays a crucial role in various fields. To address the challenges in fine-grained image retrieval tasks, this paper proposes a novel fine-grained hashing image retrieval method based on Vision Transformer (ViT) and Large Language Model (LLM). By integrating visual features with the output from the language model, the proposed method generates more robust and semantically meaningful hash codes. Experimental results demonstrate that the proposed method significantly outperforms existing methods on the CUB-200-2011 dataset, especially maintaining high retrieval accuracy with shorter hash code lengths. This approach provides an efficient and generalizable solution for fine-grained image retrieval tasks.