MFVSR: A Multi-Style Benchmark with 3,000 Videos and 16 Languages for Video Subtitle Removal
Xiyan Huang · hf · 2026-10-16
MFVSR introduces a large-scale benchmark for video subtitle removal, built on 3,000 source videos (real-HQ, real-LQ, AI-generated) with a controlled subtitle rendering engine covering 16 languages and 24 style cells, plus 953 in-the-wild videos from Douyin, Xiaohongshu, bilibili, Kuaishou and YouTube. Evaluation uses blind and mask-guided protocols across 5 spatial views, scoring restoration fidelity, background preservation and temporal consistency, with audit300/lite3000 protocols, robustness analysis and human validation.
More from Multimodal
- vLLM officially supports Muse Spark Open: multimodal, 1M context, agentic — vllm_project · 2026-10-16
- InfoVis: 9B unified model tops open-source slide generation and editing, trained on 13M records — Jiahao Wang · 2026-10-16
- RepForcing: feature-space supervision improves few-step autoregressive video distillation — zju · 2026-10-16
- SCOPE: semantic span-aware GRPO objective improves T2I prompt rewriting — university-of-rochester · 2026-10-16
- Orbia benchmark asks: do generated videos form a consistent 3D world? — tencent · 2026-10-16
- 3DCodeVerse: USC's ecosystem generates verifiable 3D worlds as code, with a 2M-sample dataset — UniversityofSouthernCalifornia · 2026-10-16