LexLIP: Lexicon-Bottlenecked Language-Image Pre-Training for Large-Scale Image-Text Retrieval

doi:doi:10.57702/00udqbf4

LexLIP: Lexicon-Bottlenecked Language-Image Pre-Training for Large-Scale Image-Text Retrieval

Image-text retrieval (ITR) is a task to retrieve the relevant images/texts, given the query from another modality. The conventional dense retrieval paradigm relies on encoding images and texts into dense representations using dual-stream encoders, however, it faces challenges with low retrieval speed in large-scale retrieval scenarios.

BibTex:

@dataset{Ziyang_Luo_and_Pu_Zhao_and_Can_Xu_and_Xiubo_Geng_and_Tao_Shen_and_Chongyang_Tao_and_Jing_Ma_and_Qingwen_Lin_and_Daxin_Jiang_2024,
    abstract = {Image-text retrieval (ITR) is a task to retrieve the relevant images/texts, given the query from another modality. The conventional dense retrieval paradigm relies on encoding images and texts into dense representations using dual-stream encoders, however, it faces challenges with low retrieval speed in large-scale retrieval scenarios.},
    author = {Ziyang Luo and Pu Zhao and Can Xu and Xiubo Geng and Tao Shen and Chongyang Tao and Jing Ma and Qingwen Lin and Daxin Jiang},
    doi = {10.57702/00udqbf4},
    institution = {No Organization},
    keyword = {'image-text retrieval', 'lexicon-weighting paradigm', 'sparse representations'},
    month = {dec},
    publisher = {TIB},
    title = {LexLIP: Lexicon-Bottlenecked Language-Image Pre-Training for Large-Scale Image-Text Retrieval},
    url = {https://service.tib.eu/ldmservice/dataset/lexlip--lexicon-bottlenecked-language-image-pre-training-for-large-scale-image-text-retrieval},
    year = {2024}
}