Research map / Video / Video-language understanding
Movie Description
Anna Rohrbach, Atousa Torabi, Marcus Rohrbach, Niket Tandon, Christopher Pal, Hugo Larochelle, Aaron C. Courville, Bernt Schiele
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 |
|---|---|---|---|
| stack IDT, LSDA and PLACES features instead of one (ablation) | Visual-Labels LSTM · MPII-MD val | METEOR: 7.10 -> 7.24 | Improved |
| assign features by label group (verbs IDT, others LSDA) (ablation) | Visual-Labels LSTM · MPII-MD val | METEOR: 7.24 -> 7.27 | Improved |
| add a places group with PLACES features (ablation) | Visual-Labels LSTM · MPII-MD val | METEOR: 7.27 -> 7.39 | Improved |
| use objects group only instead of all other labels (ablation) | Visual-Labels LSTM · MPII-MD val | METEOR: 7.39 -> 7.48 | Improved |
| restrict to labels with ROC >= 0.7 (ablation) | Visual-Labels LSTM · MPII-MD val | METEOR: 7.48 -> 7.54 | Improved |
| apply the same restriction with uniform features (ablation) | IDT+LSDA+PLACES baseline · MPII-MD val | METEOR: 7.24 -> 7.20 | No clear effect |
| use two factored LSTM layers instead of one layer (ablation) | Visual-Labels LSTM · MPII-MD val | METEOR: 7.54 -> 7.41 | Hurt |
| add dropout after the LSTM (lstm-drop) (ablation) | Visual-Labels LSTM · MPII-MD val | METEOR: 7.19 -> 7.54 | Improved |
| dropout on language only (ablation) | Visual-Labels LSTM · MPII-MD val | METEOR: 7.19 -> 7.13 | Hurt |
| dropout ratio 0.5 instead of 0.1 (ablation) | Visual-Labels LSTM · MPII-MD val | METEOR: 7.22 -> 7.54 | Improved |
| use text-labels instead of sense-labels (ablation) | SMT with IDT 30 · MPII-MD val | METEOR: 4.93 -> 5.59 | Improved |
| use Visual-Labels LSTM instead of SMT (ablation) | SMT-Best · MPII-MD test | METEOR: 5.59 -> 7.03 | Improved |
Closest papers by meaning
- AutoAD II: The Sequel - Who, When, and What in Movie Audio Description (ICCV 2023 · similarity 0.92)
- DistinctAD: Distinctive Audio Description Generation in Contexts (CVPR 2025 · similarity 0.90)
- AutoAD III: The Prequel - Back to the Pixels (CVPR 2024 · similarity 0.90)
- Contextual AD Narration with Interleaved Multimodal Sequence (CVPR 2025 · similarity 0.90)
- Evaluating Multimodal Narrative Understanding of Popular Hollywood Films (NeurIPS 2026 · similarity 0.86)
- Aligned Better, Listen Better for Audio-Visual Large Language Models (ICLR 2025 · 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.
