Research map / Video / Video-language understanding
Fully Convolutional Video Captioning with Coarse-to-Fine and Inherited Attention
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 |
|---|---|---|---|
| replace LSTM with fully convolutional decoder | video captioning model · MSVD | CIDEr: 44.1 -> 64.8 | Helped |
| replace LSTM with fully convolutional decoder | video captioning model · MSVD | training time per 10M parameters: 203sec -> 186sec | Helped |
| add coarse-to-fine and inherited attention (ablation) | FCVC · MSVD | CIDEr: 64.8 -> 79.8 | Helped |
| add coarse-to-fine and inherited attention (ablation) | FCVC · MSVD | training time per 10M parameters: 186sec -> 253sec | Hurt |
| add temporal attention in every layer (ablation) | FCVC · MSVD | CIDEr: 64.8 -> 70.7 | Helped |
| add inherited attention in every layer (ablation) | FCVC · MSVD | CIDEr: 64.8 -> 73.7 | Helped |
| use inherited attention only in last two layers instead of temporal plus inherited (ablation) | FCVC-all · MSVD | CIDEr: 77.2 -> 79.8 | Helped |
| use temporal attention in first two layers and inherited in last two (final) vs temporal in all layers (ablation) | FCVC-TA · MSVD | CIDEr: 70.7 -> 79.8 | Helped |
| fully convolutional model with attention vs best prior methods | FCVC-CF&IA · MSVD | BLEU-4: 52.3 -> 53.1 | Helped |
| fully convolutional model with attention vs TSL | FCVC-CF&IA · MSVD | CIDEr: 74.9 -> 79.8 | Helped |
Closest papers by meaning
- Non-Autoregressive Coarse-to-Fine Video Captioning (AAAI 2021 · similarity 0.90)
- Convolutional Image Captioning (CVPR 2018 · similarity 0.90)
- Distilling Vision-Language Models on Millions of Videos (CVPR 2024 · similarity 0.88)
- Fine-Grained Captioning of Long Videos through Scene Graph Consolidation (ICML 2025 · similarity 0.87)
- CALVIN: Improved Contextual Video Captioning via Instruction Tuning (NeurIPS 2024 · similarity 0.87)
- AuroraCap: Efficient, Performant Video Detailed Captioning and a New Benchmark (ICLR 2025 · similarity 0.87)
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.