Search Results for author:
Found 3 papers, 3 papers with code
Movie101v2: Improved Movie Narration Benchmark
Automatic movie narration targets at creating video-aligned plot descriptions to assist visually impaired audiences.
Less is More: Mitigating Multimodal Hallucination from an EOS Decision Perspective
In this paper, we explore a new angle of this issue: overly detailed training data hinders the model's ability to timely terminate generation, leading to continued outputs beyond visual perception limits.
Movie101: A New Movie Understanding Benchmark
Closer to real scenarios, the Movie Clip Narrating (MCN) task in our benchmark asks models to generate role-aware narration paragraphs for complete movie clips where no actors are speaking.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
