Research map / Video / Video-language understanding
Unifying Event Detection and Captioning as Sequence Generation via Pre-training
Qi Zhang, Yuqing Song, Qin Jin
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 MLM pre-training (ablation) | captioning model w/o pre-training · ActivityNet Captions val (GT events, ablation) | CIDEr: 47.7 -> 49.05 | Improved |
| add MLM pre-training (ablation) | captioning model w/o pre-training · ActivityNet Captions val (GT events, ablation) | BLEU@4: 2.22 -> 2.17 | Hurt |
| add MVFR pre-training on top of MLM (ablation) | captioning model · ActivityNet Captions val (GT events, ablation) | CIDEr: 49.05 -> 51.08 | Improved |
| add MEFM pre-training (ablation) | captioning model with MLM+MVFR · ActivityNet Captions val (GT events, ablation) | CIDEr: 51.08 -> 52.13 | Improved |
| add MEFM pre-training (ablation) | captioning model with MLM+MVFR · ActivityNet Captions val (GT events, ablation) | BLEU@4: 2.38 -> 2.59 | Improved |
| add semantic concept (CPT) features (ablation) | captioning model · ActivityNet Captions val (GT events, ablation) | CIDEr: 52.13 -> 52.42 | Improved |
| add MEFM pre-training (ablation) | detection model w/o pre-training · ActivityNet Captions val (event detection, ablation) | average recall: 54.94 -> 56.49 | Improved |
| add MLM pre-training (ablation) | detection model with MEFM · ActivityNet Captions val (event detection, ablation) | average recall: 56.49 -> 58.00 | Improved |
| add MVFR pre-training (ablation) | detection model with MEFM+MLM · ActivityNet Captions val (event detection, ablation) | average recall: 58.00 -> 58.32 | Improved |
| add MLM and MVFR tasks (ablation) | detection model with MEFM · ActivityNet Captions val (event detection, ablation) | average precision: 58.12 -> 59.95 | Improved |
| add semantic concept (CPT) features (ablation) | detection model · ActivityNet Captions val (event detection, ablation) | average recall: 58.32 -> 59.00 | Improved |
| add semantic concept (CPT) features (ablation) | detection model · ActivityNet Captions val (event detection, ablation) | average precision: 59.95 -> 60.32 | Improved |
Closest papers by meaning
- Event-Equalized Dense Video Captioning (CVPR 2025 · similarity 0.92)
- Exploring Temporal Event Cues for Dense Video Captioning in Cyclic Co-Learning (AAAI 2025 · similarity 0.89)
- Do You Remember? Dense Video Captioning with Cross-Modal Memory Retrieval (CVPR 2024 · similarity 0.87)
- Implicit Location-Caption Alignment via Complementary Masking for Weakly-Supervised Dense Video Captioning (AAAI 2025 · similarity 0.87)
- Progress-Aware Video Frame Captioning (CVPR 2025 · similarity 0.87)
- Dense Video Object Captioning from Disjoint Supervision (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.
