Research map / Vision-language and multimodal / Visual question answering
Towards Visual-Prompt Temporal Answer Grounding in Instructional Video
Shutao Li A, Bin Li, Bin Sun, Yixuan Weng
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 visual prompt (highlight features) (ablation) | VPTSL · MedVidQA test | mIoU: 55.41 -> 57.81 | Improved |
| add highlight loss (ablation) | VPTSL · MedVidQA test | mIoU: 57.44 -> 57.81 | Improved |
| add pre-trained language model weights (ablation) | VPTSL · MedVidQA test | mIoU: 18.83 -> 57.81 | Improved |
| add visual prompt (highlight features) (ablation) | VPTSL · MedVidQA test | R@1 IoU=0.3: 70.97 -> 77.42 | Improved |
| add highlight loss (ablation) | VPTSL · MedVidQA test | R@1 IoU=0.7: 43.87 -> 44.52 | Improved |
| use PLM text features instead of Word2vec (ablation) | VSLBase · MedVidQA test | mIoU: 20.15 -> 20.84 | Improved |
| use PLM text features instead of Word2vec (ablation) | VSLNet · MedVidQA test | mIoU: 20.12 -> 22.23 | Improved |
| use text span predictor over subtitles instead of visual span predictor | VPTSL vs RaNet · MedVidQA test | mIoU: 29.45 -> 57.81 | Improved |
| add subtitles to query | VSLNet · MedVidQA test | mIoU: 22.23 -> 22.61 | Improved |
| add subtitles to query | RaNet · MedVidQA test | mIoU: 27.48 -> 29.45 | Improved |
| add subtitles to query | TMLGA · MedVidQA test | mIoU: 19.80 -> 20.49 | Improved |
| increase extend rate alpha beyond 0.25 | VPTSL · MedVidQA test | mIoU | Hurt |
Closest papers by meaning
- Natural Language Video Localization: A Revisit in Span-based Question Answering Framework (TPAMI 2021 · similarity 0.90)
- CACR: Reinforcing Temporal Answer Grounding in Instructional Video via Candidate-Aware Causal Reasoning (ICML 2026 · similarity 0.89)
- Self-Chained Image-Language Model for Video Localization and Question Answering (NeurIPS 2023 · similarity 0.89)
- Track the Answer: Extending TextVQA from Image to Video with Spatio-Temporal Clues (AAAI 2025 · similarity 0.88)
- Universal Video Temporal Grounding with Generative Multi-modal Large Language Models (NeurIPS 2025 · similarity 0.87)
- Language-aware Visual Semantic Distillation for Video Question Answering (CVPR 2024 · 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.
