Research map / Video / Video-language understanding
Vinci2: Providing Proactive Assistance in Continuous Egocentric Videos
Sitong Gong, Tianyu Yan, Caixin Kang, Bo Zheng, Xiang Ruan, Huchuan Lu, Kaipeng Zhang, Yoichi Sato, Yifei Huang
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 EgoMemo memory-augmented retrieval to a VLM/LLM agent | GPT-5-mini agent · EgoServe | overall F1: 4.7 -> 8.0 | Improved |
| add multi-scale (clip/activity/session) temporal summaries (ablation) | EgoMemo · EgoServe | overall F1: 7.0 -> 8.0 | Improved |
| add VLM-based caption reconstruction (ablation) | EgoMemo · EgoServe | overall F1: 6.6 -> 8.0 | Improved |
| add visual embedding archive retrieval (ablation) | EgoMemo · EgoServe | overall F1: 6.9 -> 8.0 | Improved |
| add knowledge graph retrieval (ablation) | EgoMemo · EgoServe | overall F1: 6.5 -> 8.0 | Improved |
| add multi-scale temporal retrieval (ablation) | EgoMemo · EgoServe | overall F1: 6.8 -> 8.0 | Improved |
| add knowledge graph retrieval (ablation) | EgoMemo · EgoServe (Memory Link) | F1: 0.0 -> 4.9 | Improved |
| training-free EgoMemo vs trained EyeWO on implicit proactive tasks | EgoMemo · ESTP-Bench (implicit, All) | score: 52.5 -> 34.7 | Hurt |
| EgoMemo vs EyeWO on explicit proactive tasks | EgoMemo · ESTP-Bench (explicit, All) | score: 23.6 -> 27.6 | Improved |
| EgoMemo vs GPT-4o on backward tracing | EgoMemo · OVO-Bench (backward tracing) | average accuracy: 60.75 -> 49.60 | Hurt |
| EgoMemo vs GPT-4o on real-time perception | EgoMemo · OVO-Bench (real-time perception) | average accuracy: 64.46 -> 75.15 | Improved |
| EgoMemo vs EgoThinker on EgoSchema | EgoMemo · EgoSchema | accuracy: 67.6 -> 74.8 | 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.
- EgoThinker — EgoThinker: Unveiling Egocentric Reasoning with Spatio-Temporal CoT (NeurIPS 2025) · better on 2 of 3 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
- EgoExoBench: A Benchmark for First- and Third-person View Video Understanding in MLLMs (NeurIPS 2025 · similarity 0.89)
- EgoLife: Towards Egocentric Life Assistant (CVPR 2025 · similarity 0.89)
- In the Eye of MLLM: Benchmarking Egocentric Video Intent Understanding with Gaze-Guided Prompting (NeurIPS 2025 · similarity 0.88)
- Omnia de EgoTempo: Benchmarking Temporal Understanding of Multi-Modal LLMs in Egocentric Videos (CVPR 2025 · similarity 0.88)
- EgoThinker: Unveiling Egocentric Reasoning with Spatio-Temporal CoT (NeurIPS 2025 · similarity 0.88)
- Ego2Web: A Web Agent Benchmark Grounded in Egocentric Videos (CVPR 2026 · 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.
