Research map / Video / Video-language understanding
Move Forward and Tell: A Progressive Generator of Video Descriptions
Yilei Xiong, Bo Dai, Dahua Lin
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 |
|---|---|---|---|
| subsample negatives (S3) instead of complete sequences when training event selection (ablation) | event selection network · ActivityNet Captions (paragraph, test half of val) | CIDEr: 9.25 -> 14.15 | Improved |
| subsample negatives (S3) instead of subsampling at intervals (ablation) | event selection network · ActivityNet Captions (paragraph, test half of val) | CIDEr: 8.04 -> 14.15 | Improved |
| subsample negatives (S3) instead of complete sequences (ablation) | event selection network · ActivityNet Captions (paragraph, test half of val) | Repetition Evaluation (RE): 26.11 -> 17.59 | Improved |
| add self-critical reinforcement learning with sentence and paragraph CIDEr rewards (ablation) | caption generation network · ActivityNet Captions (paragraph, test half of val) | CIDEr: 12.81 -> 14.15 | Improved |
| add self-critical reinforcement learning (ablation) | caption generation network · ActivityNet Captions (paragraph, test half of val) | Repetition Evaluation (RE): 19.10 -> 17.59 | Improved |
| add self-critical reinforcement learning (ablation) | caption generation network · ActivityNet Captions (paragraph, test half of val) | BLEU@4: 8.53 -> 8.45 | No clear effect |
| add range feature to visual feature for event selection (ablation) | event selection network · ActivityNet Captions (paragraph, test half of val) | CIDEr: 10.19 -> 10.54 | Improved |
| add caption feature to visual feature for event selection (ablation) | event selection network · ActivityNet Captions (paragraph, test half of val) | CIDEr: 10.19 -> 11.29 | Improved |
| add caption feature to visual + range features (ablation) | event selection network · ActivityNet Captions (paragraph, test half of val) | CIDEr: 10.54 -> 14.15 | Improved |
| add range feature to visual + caption features (ablation) | event selection network · ActivityNet Captions (paragraph, test half of val) | CIDEr: 11.29 -> 14.15 | Improved |
| add caption feature to visual feature (ablation) | event selection network · ActivityNet Captions (paragraph, test half of val) | Repetition Evaluation (RE): 29.00 -> 18.92 | Improved |
| select events progressively (MFT) instead of NMS over dense captions | paragraph generator · ActivityNet Captions (paragraph, test half of val) | CIDEr: 3.52 -> 14.15 | Improved |
Closest papers by meaning
- Video Paragraph Captioning Using Hierarchical Recurrent Neural Networks (CVPR 2016 · similarity 0.91)
- Towards Diverse Paragraph Captioning for Untrimmed Videos (CVPR 2021 · similarity 0.90)
- Sketch, Ground, and Refine: Top-Down Dense Video Captioning (CVPR 2021 · similarity 0.89)
- Video ReCap: Recursive Captioning of Hour-Long Videos (CVPR 2024 · similarity 0.88)
- Progress-Aware Video Frame Captioning (CVPR 2025 · similarity 0.87)
- OmniViD: A Generative Framework for Universal Video Understanding (CVPR 2024 · 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.
