Abstract: This paper proposes a novel audio and visual class association method based on contrastive learning that can obtain not only one-to-one but also one-to-many, many-to-many, and even no ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback