Research map / Video / Video-language understanding
EgoEnv: Human-centric environment representations from egocentric video
Tushar Nagarajan, Santhosh Kumar Ramakrishnan, Ruta Desai, James Hillis, Kristen Grauman
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 EgoEnv features to VSLNet | VSLNet · NLQ MP3D | Rank1 avg: 28.26 -> 32.52 | Improved |
| add EgoEnv features to VSLNet | VSLNet · NLQ HouseTours | Rank1 avg: 35.31 -> 43.08 | Improved |
| add EgoEnv features to VSLNet | VSLNet · NLQ Ego4D val | Rank1 avg: 4.29 -> 4.77 | Improved |
| EgoEnv features instead of EgoTopo video-level features | VSLNet · NLQ HouseTours | Rank1 avg: 35.66 -> 43.08 | Improved |
| replace MAE self-supervised features with EgoEnv | VSLNet · NLQ MP3D | Rank1 avg: 29.62 -> 32.52 | Improved |
| EgoEnv vs EPC with ground-truth pose at inference | VSLNet · NLQ MP3D | Rank1@0.5: 27.64 -> 26.85 | Hurt |
| EPC on HouseTours with noisy estimated pose | EPC · NLQ HouseTours | Rank1 avg: 35.66 -> 35.52 | No clear effect |
| replace scratch transformer with pretrained EgoEnv | EgoEnv · NLQ HouseTours | Rank1 avg: 31.43 -> 43.08 | Improved |
| augment baseline with EgoEnv features on Ego4D challenge | NaQ · Ego4D NLQ test | Rank1 avg: 17.67 -> 18.82 | Improved |
| evaluate EgoEnv on indoor-aligned Ego4D scenes | EgoEnv · Ego4D NLQ val | Rank1 avg: 20.35 -> 25.16 | Improved |
| add pose embeddings to local state prediction (ablation) | EgoEnv · local state prediction | mAP: +0.7 mAP forward; +2.2 right | Improved |
| embed ground-truth pose in input (ablation) | EgoEnv · MP3D vs HouseTours | downstream performance | Mixed |
Closest papers by meaning
- Whole-Body Conditioned Egocentric Video Prediction (NeurIPS 2025 · similarity 0.88)
- DIV-FF: Dynamic Image-Video Feature Fields For Environment Understanding in Egocentric Videos (CVPR 2025 · similarity 0.88)
- EgoGen: An Egocentric Synthetic Data Generator (CVPR 2024 · similarity 0.87)
- SAW-Bench: Learning Situated Awareness in the Real World (ICML 2026 · similarity 0.87)
- Retrieval-Augmented Egocentric Video Captioning (CVPR 2024 · similarity 0.87)
- Learning to Evolve Scenes: Reasoning about Human Activities with Scene Graphs (NeurIPS 2026 · similarity 0.86)
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.
