Abstract: Current cross-modal retrieval methods heavily rely on accurate semantic labels or sample similarity measurements, and need to search for the nearest samples among all samples in the huge ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback