Research map / Video / Video generation
LOVE: Benchmarking and Evaluating Text-to-Video Generation and Video-to-Text Interpretation
Jiarui Wang, Huiyu Duan, Ziheng Jia, Zicheng Zhang, Yu Zhao, Juntong Wang, Guangtao Zhai, Xiongkuo Min
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 temporal features (full LOVE vs. row 5 variant) (ablation) | LOVE · AIGVE-60K perception | SRCC: 0.734 -> 0.793 | Improved |
| add text-defined quality-level initialization (ablation) | LOVE · AIGVE-60K perception | SRCC: 0.642 -> 0.734 | Improved |
| add LoRA fine-tuning on vision encoder and LLM (ablation) | LOVE · AIGVE-60K QA | accuracy: 65.8% -> 77.3% | Improved |
| fine-tune LMM on AIGVE-60K instead of zero-shot (InternVL3 9B) | InternVL3-9B · AIGVE-60K perception instance-level | SRCC: 0.2731 -> 0.6421 | Improved |
| use LOVE reward (perception + correspondence) vs. perception only (ablation) | Wan2.1 GRPO · HPSv2 evaluation | HPSv2: 1.6132 -> 1.6250 | Improved |
| use LOVE-Reward with GRPO on Wan2.1 | Wan2.1 · LOVE correspondence evaluation | LOVE correspondence score: 0.5175 -> 0.5268 | Improved |
| use CLIPScore as GRPO reward instead of LOVE | Wan2.1 GRPO · HPSv2 evaluation | HPSv2: 1.5576 -> 1.5576 | No clear effect |
| use HPSv2 or ImageReward as GRPO reward, evaluated on LOVE perception | Wan2.1 GRPO · LOVE perception evaluation | LOVE perception score: 0.5118 -> 0.5091 (HPSv2+GRPO) | Hurt |
| train on AIGVE-60K and test zero-shot on other datasets vs. VQAScore trained on GenAI-Bench | LOVE · GenAI-Bench | SRCC: 0.632 -> 0.682 | Improved |
| use an LMM-based evaluator vs. handcrafted metrics | AIGV evaluation · AIGVE-60K T2V correspondence (model-level) | SRCC: 0.7321 (BMPRI) -> 0.9778 (LOVE) | Improved |
Closest papers by meaning
- Neuro-Symbolic Evaluation of Text-to-Video Models using Formal Verification (CVPR 2025 · similarity 0.89)
- Boosting Text-to-Video Generative Model with MLLMs Feedback (NeurIPS 2024 · similarity 0.88)
- T2V-CompBench: A Comprehensive Benchmark for Compositional Text-to-video Generation (CVPR 2025 · similarity 0.88)
- AIGV-Assessor: Benchmarking and Evaluating the Perceptual Quality of Text-to-Video Generation with LMM (CVPR 2025 · similarity 0.88)
- Rethinking Human Evaluation Protocol for Text-to-Video Models: Enhancing Reliability, Reproducibility, and Practicality (NeurIPS 2024 · similarity 0.88)
- Q-Bench-Video: Benchmark the Video Quality Understanding of LMMs (CVPR 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.
