Research map / Video / Video-language understanding
Semantic Grouping Network for Video Captioning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use semantic groups aligned to phrases from the partially decoded caption instead of frame-level temporal attention (ablation) | TA baseline · MSR-VTT | CIDEr-D: 42.2 -> 47.9 | Helped |
| Add phrase suppression to remove similar candidate phrases (ablation) | TA + Semantic Aligner · MSR-VTT | CIDEr-D: 47.9 -> 48.5 | Helped |
| Add Contrastive Attention loss to improve phrase-frame alignment (ablation) | TA + Semantic Aligner + Phrase Suppressor · MSR-VTT | CIDEr-D: 48.5 -> 49.5 | Helped |
| Form semantic groups using single words instead of phrases, omitting Phrase Encoder and Phrase Suppressor (ablation) | SGN · MSR-VTT | CIDEr-D: 49.5 -> 46.6 | Hurt |
| Use SGN rather than ORG-TRL | Video captioning on MSR-VTT · MSR-VTT | CIDEr-D: 50.9 -> 49.5 | Hurt |
| Use SGN rather than MARN | Video captioning on MSVD · MSVD | CIDEr-D: 92.2 -> 94.3 | Helped |
| Use semantic grouping with repeated phrase-conditioned alignment instead of direct temporal attention | Video captioning inference · MSVD | videos per second: 865 -> 657 | Hurt |
| Use semantic grouping with repeated phrase-conditioned alignment instead of direct temporal attention | Video captioning inference · MSR-VTT | videos per second: 268 -> 203 | Hurt |
Closest papers by meaning
- Semantic Compositional Networks for Visual Captioning (CVPR 2017 · similarity 0.87)
- Memory-Attended Recurrent Network for Video Captioning (CVPR 2019 · similarity 0.87)
- Hierarchical Modular Network for Video Captioning (CVPR 2022 · similarity 0.86)
- Fine-Grained Captioning of Long Videos through Scene Graph Consolidation (ICML 2025 · similarity 0.86)
- VideoCon: Robust Video-Language Alignment via Contrast Captions (CVPR 2024 · similarity 0.84)
- VidLA: Video-Language Alignment at Scale (CVPR 2024 · similarity 0.84)
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.