Abstract: To establish semantic associations between images and texts, existing Image-Text Retrieval (ITR) methods primarily focus on fixed-scale fragments, which only identify explicit semantic ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results