Research map / Video / Video segmentation and tracking
PR-Net: Preference Reasoning for Personalized Video Highlight Detection
Runnan Chen, Penghao Zhou, Wenzhe Wang, Nenglun Chen, Pai Peng, Xing Sun, Wenping Wang
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 the U-Net style contextual encoder instead of a residual block (ablation) | PR-Net · PHD-GIFs | mAP (%): 16.69 -> 18.66 | Improved |
| use bi-directional contrastive loss instead of ranking loss (ablation) | PR-Net · PHD-GIFs | mAP (%): 16.85 -> 18.66 | Improved |
| use bi-directional contrastive loss instead of cross-entropy loss (ablation) | PR-Net · PHD-GIFs | mAP (%): 17.09 -> 18.66 | Improved |
| use bi-directional contrastive loss instead of triplet loss (ablation) | PR-Net · PHD-GIFs | mAP (%): 17.26 -> 18.66 | Improved |
| add user-specific preference (ablation) | PR-Net · PHD-GIFs | mAP (%): 16.78 -> 18.66 | Improved |
| use attention-based user-specific preference instead of mean history embedding (ablation) | PR-Net · PHD-GIFs | mAP (%): 16.87 -> 18.66 | Improved |
| add generic preference to the user-specific preference (ablation) | PR-Net · PHD-GIFs | mAP (%): 17.57 -> 18.66 | Improved |
| add generic preference to mean-history variant (ablation) | PR-Net · PHD-GIFs | mAP (%): 16.87 -> 17.21 | Improved |
| provide up to 5 history highlights instead of none (ablation) | PR-Net · PHD-GIFs | mAP (%): 16.78 -> 17.48 | Improved |
| use all available history instead of limiting it to 5 (ablation) | PR-Net · PHD-GIFs | mAP (%): 17.48 -> 18.66 | Improved |
| use PR-Net instead of A-VHD | personalized highlight detection · PHD-GIFs | mAP (%): 16.68 -> 18.66 | Improved |
| use PR-Net instead of PHD-GIFs | personalized highlight detection · PHD-GIFs | mAP (%): 16.25 -> 18.66 | Improved |
Closest papers by meaning
- PAC-Net: Highlight Your Video via History Preference Modeling (ECCV 2022 · similarity 0.92)
- Adaptive Video Highlight Detection by Learning from User History (ECCV 2020 · similarity 0.89)
- Contrastive Learning for Unsupervised Video Highlight Detection (CVPR 2022 · similarity 0.89)
- Bridging the Gap: A Unified Video Comprehension Framework for Moment Retrieval and Highlight Detection (CVPR 2024 · similarity 0.86)
- Watch Video, Catch Keyword: Context-aware Keyword Attention for Moment Retrieval and Highlight Detection (AAAI 2025 · similarity 0.85)
- ShotVL: Human-Centric Highlight Frame Retrieval via Language Queries (AAAI 2025 · similarity 0.83)
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.
