Research map / Video / Video segmentation and tracking
Deep Reinforcement Learning for Unsupervised Video Summarization With Diversity-Representativeness Reward
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 |
|---|---|---|---|
| combine diversity and representativeness rewards instead of using either alone (ablation) | DSN · SumMe | F-score: 40.5 -> 41.4; 40.7 -> 41.4 | Helped |
| combine diversity and representativeness rewards instead of using either alone (ablation) | DSN · TVSum | F-score: 56.2 -> 57.6; 56.9 -> 57.6 | Helped |
| remove temporal-distance lambda technique (ablation) | D-DSN · SumMe and TVSum | F-score: 39.3 -> 40.5; 55.7 -> 56.2 | Hurt |
| add supervised MLE objective (ablation) | DR-DSN · SumMe | F-score: 41.4 -> 42.1; 1.7% improvements | Helped |
| add supervised MLE objective (ablation) | DR-DSN · TVSum | F-score: 57.6 -> 58.1; 0.9% improvements | Helped |
| replace LSTM with GRU (ablation) | DR-DSN · SumMe | F-score: 41.4 -> 41.2 | Hurt |
| replace LSTM with GRU (ablation) | DR-DSN · TVSum | F-score: 57.6 -> 56.7 | Hurt |
| replace LSTM with GRU (ablation) | DR-DSN_sup · SumMe and TVSum | F-score: 42.1 -> 41.5; 58.1 -> 57.8 | Hurt |
| train with the diversity-representativeness RL reward rather than supervised cross-entropy alone | DSN · SumMe and TVSum | F-score: 38.2 -> 41.4; 54.5 -> 57.6 | Helped |
| use DR-DSN instead of GAN_dpp | video summarization · SumMe and TVSum, canonical | F-score: 39.1 -> 41.4; 51.7 -> 57.6 | Helped |
| use DR-DSN_sup instead of GAN_sup | video summarization · SumMe and TVSum, canonical | F-score: 41.7 -> 42.1; 56.3 -> 58.1 | Helped |
| use DR-DSN_sup instead of GAN_sup | video summarization · TVSum, augmented | F-score: 61.2 vs. 59.8 | Hurt |
Closest papers by meaning
- Video Summarization by Learning From Unpaired Data (CVPR 2019 · similarity 0.90)
- Video Summarization with Large Language Models (CVPR 2025 · similarity 0.86)
- Video Summarization Using Denoising Diffusion Probabilistic Model (AAAI 2025 · similarity 0.86)
- V2Xum-LLM: Cross-Modal Video Summarization with Temporal Prompt Instruction Tuning (AAAI 2025 · similarity 0.85)
- Scaling Up Video Summarization Pretraining with Large Language Models (CVPR 2024 · similarity 0.85)
- Reinforcing Structured Chain-of-Thought for Video Understanding (CVPR 2026 · similarity 0.85)
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.