Research map / Video / Video-language understanding
GuideMe: Benchmarking Multi-Domain Task Guidance and Intervention in Streaming Video
FANG LIU, Jinpeng Chen, Ke Xu, Yuhao LIU, Huankang Guan, Xudong LU, Yang Bo, Gerhard P. Hancke, Rui Liu, Rynson Lau
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 |
|---|---|---|---|
| evaluate on error events instead of instruction events | Gemini 3 Pro · GuideMe-Test | sF1: 42.0 -> 33.8 | Hurt |
| evaluate on error events instead of instruction events | Doubao-Seed-1.8 · GuideMe-Test | sF1: 49.6 -> 41.9 | Hurt |
| fine-tune Qwen3-VL-8B on GuideMe train split | Qwen3-VL-8B · GuideMe-Test | sF1: 35.7 -> 39.7 | Improved |
| fine-tune Qwen3-VL-8B on GuideMe train split | Qwen3-VL-8B · GuideMe-Test | Partly Correct rate: 50.5 -> 35.3 | Hurt |
| fine-tune Qwen3-VL-8B on GuideMe train split | Qwen3-VL-8B · GuideMe-Test | behavior Score: 31.9 -> 42.1 | Improved |
| scale Qwen3-VL from 8B to 30B-A3B | Qwen3-VL · GuideMe-Test | sF1: 35.7 -> 35.6 | No clear effect |
| scale to Qwen3.5-397B-A17B | Qwen3-VL · GuideMe-Test | sF1: 35.6 -> 21.7 | Hurt |
| use dense 1 s sampling instead of anchor-based sampling (ablation) | Doubao-Seed-1.8 · GuideMe 100-video subset | behavior Score: 38.6 -> 90.3 | Mixed |
| replace GT dialogue history with the model's own predictions (ablation) | Doubao-Seed-1.8 · GuideMe 100-video subset | sF1: 36.9 -> 20.0 | Hurt |
| shrink sliding window from 60 s to 30 s (ablation) | Gemini 3 Pro · anchor sampling | behavior Score: 44.8 -> 40.3 | Hurt |
| enlarge sliding window from 60 s to 120 s (ablation) | Gemini 3 Pro · anchor sampling | behavior Score: 44.8 -> 47.6 | Improved |
| use 2 s sampling interval instead of 1 s (ablation) | Gemini 3 Pro · dense sampling | behavior Score: 47.9 -> 40.8 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 on most shared results.
- Dispider — Dispider: Enabling Video LLMs with Active Real-Time Interaction via Disentangled Perception, Decision, and Reaction (CVPR 2025) · better on 1 of 2 shared results
- LiveStar — LiveStar: Live Streaming Assistant for Real-World Online Video Understanding (NeurIPS 2025) · better on 2 of 2 shared results
- VideoLLM-online — VideoLLM-online: Online Video Large Language Model for Streaming Video (CVPR 2024) · better on 2 of 2 shared results
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
- Can Multi-Modal LLMs Provide Live Step-by-Step Task Guidance? (NeurIPS 2025 · similarity 0.89)
- Needle In A Video Haystack: A Scalable Synthetic Evaluator for Video MLLMs (ICLR 2025 · similarity 0.87)
- SLVMBench: Skill Learning from Video Memory (NeurIPS 2026 · similarity 0.87)
- MMOU: A Massive Multi-Task Omni Understanding and Reasoning Benchmark for Long and Complex Real-World Videos (NeurIPS 2026 · similarity 0.87)
- E.T. Bench: Towards Open-Ended Event-Level Video-Language Understanding (NeurIPS 2024 · similarity 0.87)
- Unleashing Hour-Scale Video Training for Long Video-Language Understanding (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.
