24 DAILY WACV Sunday Poster Presentation To what extent can multimodal cues replace video? The main goal of Manuel’s work is to study whether the information contained in a single frame is enough to substitute video aggregation in some context. “We first analyze what a single frame can provide,” Manuel explains “and we study also how multimodal cues can enhance the information provided by this single frame. In particular, we study how depth information can improve this frame and how long-term context extracted by either a visual language model or the action history by previous observations can contribute to enhance the next action anticipation.” Action Anticipation at a Glimpse: To What Extent Can Multimodal Cues Replace Video? Manuel Benavent-Lledo is currently a postdoc researcher at the University of Alicante. He speaks to us ahead of his poster presentation later today.
RkJQdWJsaXNoZXIy NTc3NzU=