A multiple cross-attention network is proposed to facilitate query-video and query-subtitle matching. Performance improvement in video-subtitle moment ...

Apr 1, 2022 · A multiple cross-attention network is proposed to facilitate query-video and query-subtitle matching. •. Performance improvement ...

MMCDA contains three parts: the shared feature encoders to extract the video and query features in each domain, the cross-modal attention to capture the video- ...

The video moment retrieval task aims to fetch a target moment in an untrimmed video, which best matches the semantics of a sentence query.

Video Moment Retrieval (VMR) aims to retrieve temporal segments in untrimmed videos corresponding to a given language query by constructing cross-modal ...

through a cross-transformer encoder with two layers. The cross-attention between video and text embeddings can be formulated as: CrossAtt(Qv,Kt,Vt) ...

Illustration of the intrinsic characteristics of Moment. Retrieval and Highlight Detection. We visualize the attention map of the same video under two tasks.

Therefore, our proposed two attention sub-networks can recognize the most relevant objects and interactions in the video, and simultaneously highlight the ...

In particular, we design a memory attention mechanism to emphasize the visual features mentioned in the query and simultaneously incorporate their context. In ...

Oct 28, 2024 · Our approach introduces a new retention mechanism into the multimodal Transformer architecture, incorporating modality-specific attention modes.