Research map / Vision-language and multimodal / Visual question answering
Ego-Grounding for Personalized Question-Answering in Egocentric Videos
Junbin Xiao, Shenglang Zhang, Pengxiang Zhu, Angela Yao
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 |
|---|---|---|---|
| use thinking variant instead of instruct (ablation) | Qwen3-VL-8B · MyEgo OE | accuracy: 36.4 -> 34.3 | Hurt |
| use thinking variant instead of instruct (ablation) | InternVL3.5-8B · MyEgo OE | accuracy: 29.9 -> 32.4 | Improved |
| increase input frames from 32 to 128 | LongVA/LongVU · MyEgo | accuracy | No clear effect |
| scale up model parameters | MLLMs · MyEgo | accuracy | No clear effect |
| feed question and answer moment key frames instead of uniform sampling (ablation) | Gemini-2.5 Pro · MyEgo OE (Previous) | accuracy: 40.3 -> 51.5 | Improved |
| feed question and answer moment key frames instead of uniform sampling (ablation) | Qwen2.5-VL-7B · MyEgo OE (Previous) | accuracy: 33.2 -> 42.4 | Improved |
| feed question and answer moment key frames instead of uniform sampling (ablation) | Qwen3-VL-8B-Instruct · MyEgo OE (Current) | accuracy: 36.9 -> 46.3 | Improved |
| increase uniform frames from 8 to 64 (ablation) | InternVL3-8B · MyEgo OE | accuracy: 28.7 -> 34.7 (peaks 16 frames) | No clear effect |
| sample 8 frames backward from question moment instead of 1 (ablation) | InternVL3-8B · MyEgo MC | accuracy: +6.8% | Improved |
| replace first-person pronouns with 'the camera wearer' (ablation) | LLaVA-Video · MyEgo OE | accuracy: 34.5 -> 35.9 | Improved |
| replace first-person pronouns with 'the camera wearer' (ablation) | InternVL3.5-8B · MyEgo OE | accuracy: 33.1 -> 33.2 | No clear effect |
| remove personalization cues from prompt (ablation) | InternVL3.5-8B · MyEgo OE | accuracy: 33.1 -> 31.7 | Hurt |
Closest papers by meaning
- Omnia de EgoTempo: Benchmarking Temporal Understanding of Multi-Modal LLMs in Egocentric Videos (CVPR 2025 · similarity 0.92)
- Grounded Question-Answering in Long Egocentric Videos (CVPR 2024 · similarity 0.91)
- MMEgo: Towards Building Egocentric Multimodal LLMs for Video QA (ICLR 2025 · similarity 0.91)
- EgoTextVQA: Towards Egocentric Scene-Text Aware Video Question Answering (CVPR 2025 · similarity 0.89)
- In the Eye of MLLM: Benchmarking Egocentric Video Intent Understanding with Gaze-Guided Prompting (NeurIPS 2025 · similarity 0.89)
- EgoLife: Towards Egocentric Life Assistant (CVPR 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.
