Jang, Y. K., Kim, D. (2025). Visual delta generation with large multi-modal models enhances composed image retrieval using unlabeled data. Scientific Reports. https://doi.org/10.1038/s41598-025-07798-6