Research map / Video / Video-language understanding
VidChapters-7M: Video Chapters at Scale
Antoine Yang, Arsha Nagrani, Ivan Laptev, Josef Sivic, Cordelia Schmid
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add visual input to speech-only model (ablation) | Vid2Seq · VidChapters-7M chapter generation (test) | SODA_c: 10.5 -> 11.4 | Improved |
| replace speech input with visual-only input (ablation) | Vid2Seq · VidChapters-7M chapter generation (test) | SODA_c: 10.5 -> 5.5 | Hurt |
| add HowTo100M pretraining (speech+visual) (ablation) | Vid2Seq · VidChapters-7M chapter generation (test) | SODA_c: 10.6 -> 11.4 | Improved |
| add HowTo100M pretraining (speech only) (ablation) | Vid2Seq · VidChapters-7M chapter generation (test) | SODA_c: 10.2 -> 10.5 | Improved |
| add HowTo100M pretraining (visual only) (ablation) | Vid2Seq · VidChapters-7M chapter generation (test) | SODA_c: 3.1 -> 5.5 | Improved |
| add visual input to speech-only model (ablation) | Vid2Seq · VidChapters-7M chapter title generation given GT boundaries (test) | CIDEr: 105.3 -> 120.5 | Improved |
| add HowTo100M pretraining (ablation) | Vid2Seq · VidChapters-7M chapter title generation given GT boundaries (test) | CIDEr: 110.8 -> 120.5 | Improved |
| add VidChapters-7M (ASR+Chap.) pretraining on top of C4+HTM (ablation) | Vid2Seq · YouCook2 (val) finetuned | CIDEr: 53.2 -> 67.2 | Improved |
| add VidChapters-7M pretraining | PDVC · YouCook2 (val) finetuned | CIDEr: 28.8 -> 34.7 | Improved |
| add VidChapters-7M ASR only to C4+HTM pretraining (ablation) | Vid2Seq · YouCook2 (val) finetuned | CIDEr: 53.2 -> 50.1 | Hurt |
| increase fraction of VidChapters-7M used for pretraining (1% to 10% to 100%) (ablation) | Vid2Seq · YouCook2 (val) finetuned | CIDEr: 52.7 -> 63.9 -> 67.2 | Improved |
| increase fraction of VidChapters-7M used for pretraining (1% to 10% to 100%) (ablation) | Vid2Seq · YouCook2 (val) zero-shot | CIDEr: 7.2 -> 11.5 -> 13.3 | Improved |
Closest papers by meaning
- Chapter-Llama: Efficient Chaptering in Hour-Long Videos with LLMs (CVPR 2025 · similarity 0.90)
- MovieBench: A Hierarchical Movie Level Dataset for Long Video Generation (CVPR 2025 · similarity 0.87)
- Video ReCap: Recursive Captioning of Hour-Long Videos (CVPR 2024 · similarity 0.86)
- VideoUFO: A Million-Scale User-Focused Dataset for Text-to-Video Generation (NeurIPS 2025 · similarity 0.86)
- Long Video Diffusion Generation with Segmented Cross-Attention and Content-Rich Video Data Curation (CVPR 2025 · similarity 0.86)
- Panda-70M: Captioning 70M Videos with Multiple Cross-Modality Teachers (CVPR 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
