Seeing Before Synthesizing: VLM-Guided Transition Event Discovery for Weakly-Supervised Dense Video Captioning
This work proposes Seeing Before Synthesizing (SBS), a framework that adaptively provides visually grounded linguistic guidance only where warranted and generates frame-level narratives for the inter-event gaps and detects transitions from the semantic variation across them.
Ye-Chan Kim, Seung-Hee Choi, SeungJu Cha et al.
· 0 citations