Research map / Video / Video generation
VMBench: A Benchmark for Perception-Aligned Video Motion Generation
Xinran Ling, Chen Zhu, Meiqi Wu, Hangyu Li, Xiaokun Feng, Cundian Yang, Aiming Hao, Jiashu Zhu, Jiahong Wu, Xiangxiang Chu
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 TCS (ablation) | PMM · Human preference accuracy | accuracy (%): 66.9 -> 70.6 | Improved |
| add PAS (ablation) | PMM · Human preference accuracy | accuracy (%): 68.7 -> 70.6 | Improved |
| add OIS (ablation) | PMM · Human preference accuracy | accuracy (%): 64.6 -> 70.6 | Improved |
| add MSS (ablation) | PMM · Human preference accuracy | accuracy (%): 65.2 -> 70.6 | Improved |
| add CAS (ablation) | PMM · Human preference accuracy | accuracy (%): 64.1 -> 70.6 | Improved |
| add MSS to CAS (ablation) | PMM · Human preference accuracy | accuracy (%): 58.9 -> 66.1 | Improved |
| add OIS to CAS+MSS (ablation) | PMM · Human preference accuracy | accuracy (%): 66.1 -> 67.3 | Improved |
| add PAS and TCS to CAS+MSS+OIS (ablation) | PMM · Human preference accuracy | accuracy (%): 67.3 -> 70.6 | Improved |
| use PPM instead of best MLLM prompting (InternVideo2.5) | PPM vs InternVideo2.5 · Human alignment (VMBench) | Avg Spearman rho x100: 26.9 -> 62.2 | Improved |
| use PPM instead of Dover Technical (best rule-based avg) | PPM vs Dover Technical · Human alignment (VMBench) | Avg Spearman rho x100: 20.6 -> 62.2 | Improved |
| use PPM PAS instead of RAFT PAS | PPM vs RAFT · Human alignment (VMBench) | PAS Spearman rho x100: 47.7 -> 65.2 | Improved |
| use PPM TCS instead of CLIP TCS | PPM vs CLIP · Human alignment (VMBench) | TCS Spearman rho x100: 28.0 -> 54.5 | Improved |
Closest papers by meaning
- VBench: Comprehensive Benchmark Suite for Video Generative Models (CVPR 2024 · similarity 0.92)
- Aligning Human Motion Generation with Human Perceptions (ICLR 2025 · similarity 0.90)
- Video-Bench: Human-Aligned Video Generation Benchmark (CVPR 2025 · similarity 0.90)
- MotionBench: Benchmarking and Improving Fine-grained Video Motion Understanding for Vision Language Models (CVPR 2025 · similarity 0.88)
- WorldSimBench: Towards Video Generation Models as World Simulators (ICML 2025 · similarity 0.87)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 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.
