Research map / Video / Video-language understanding
ViTexQA: A Multi-Frame Temporal Perception Dataset for Video Text Question Answering
Zhentao Guo, Chen Duan, Tongkun Guan, Zining Wang, Kai zhou, Pengfei Yan
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 ViTexQA training data (plain SFT) to Qwen3-VL | Qwen3-VL-8B · ViTexQA | ROUGE-L: 66.7 -> 76.9 (+10.2) | Improved |
| add ViTexQA training data to MiniCPM-V4.5 | MiniCPM-V4.5 · ViTexQA | ROUGE-L: 70.1 -> 77.2 (+7.1) | Improved |
| add ViTexQA training data to Qwen3-VL (out-of-domain MME-VideoOCR) | Qwen3-VL-8B · MME-VideoOCR | score: 68.5 -> 74.3 (+5.8) | Improved |
| add ViTexQA training data to Qwen3-VL (single-image TextVQA) | Qwen3-VL-8B · TextVQA | accuracy: 84.5 -> 84.6 (+0.1) | No clear effect |
| add ViTexQA training data to MiniCPM-V4.5 (Video-MME) | MiniCPM-V4.5 · Video-MME | score: 68.8 -> 70.6 (+1.6) | Hurt |
| add CoT-guided SFT on top of ViTexQA data (ablation) | FrameThinker · ViTexQA | ROUGE-L: 76.3 -> 81.4 (5.1% over No.(2)) | Improved |
| add temporally-grounded RL after CoT-SFT (ablation) | FrameThinker · ViTexQA | ROUGE-L: 81.4 -> 83.5 | Improved |
| add temporally-grounded RL after CoT-SFT (ablation) | FrameThinker · MME-VideoOCR | score: 85.3 -> 88.1 | Improved |
| add CoT-guided SFT and RL plus ViTexQA data versus the baseline (ablation) | FrameThinker · ViTexQA | ROUGE-L: 73.5 -> 83.5 (+10.0) | Improved |
| add ViTexQA data only (no CoT-SFT, no RL) to the baseline FrameThinker model (ablation) | FrameThinker · ViTexQA | ROUGE-L: 73.5 -> 76.3 (+2.8) | Improved |
| increase input resolution (ablation) | FrameThinker · ViTexQA | ROUGE-L | No clear effect |
| increase number of input frames (ablation) | FrameThinker · ViTexQA | ROUGE-L | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- LongVU — LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding (ICML 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Not Another Text Benchmark: Putting the “Visual" Back in Visual Question Answering for Large Video Models (NeurIPS 2026 · similarity 0.91)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.90)
- Language-aware Visual Semantic Distillation for Video Question Answering (CVPR 2024 · similarity 0.90)
- VideoEspresso: A Large-Scale Chain-of-Thought Dataset for Fine-Grained Video Reasoning via Core Frame Selection (CVPR 2025 · similarity 0.90)
- Temporal Reasoning Transfer from Text to Video (ICLR 2025 · similarity 0.89)
- Temporal Chain of Thought: Long-Video Understanding by Thinking in Frames (NeurIPS 2025 · similarity 0.89)
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.
