Qi-Kang Deng

Image-Based Fusion of Textual and Visual Information for Product Classification in Videos

A hybrid multimodal framework combining intermediate cross-attention fusion with subsequent late fusion is proposed, which provides the strongest overall performance.