the archive
#multimodal fusion
3 resultscs.MM2019
Exploiting Temporal Relationships in Video Moment Localization with Natural Language
Songyang Zhang, Jinsong Su, Jiebo Luo
The paper proposes a Temporal Compositional Modular Network that parses a query sentence into main, context, and temporal parts and matches each to video segments using visual and…
#video moment localization#natural language grounding#temporal reasoning#multimodal fusion
cs.CV2019
What Would You Expect? Anticipating Egocentric Actions with Rolling-Unrolling LSTMs and Modality Attention
Antonino Furnari, Giovanni Maria Farinella
The paper proposes a two‑LSTM architecture with a modality‑attention mechanism to predict future actions and objects in first‑person video, using RGB, optical flow, and object feat…
#egocentric action anticipation#multimodal fusion#lstm#modality attention
cs.CV2019
Two-Stream Video Classification with Cross-Modality Attention
Lu Chi, Guiyu Tian, Yadong Mu +1
The paper introduces a cross-modality attention operation and a CMA block to fuse information from multiple video streams more effectively than traditional score-level fusion, demo…
#video classification#cross-modality attention#two-stream networks#attention mechanisms