Research map / 3D vision / Human pose estimation
Identifying First-Person Camera Wearers in Third-Person Videos
Chenyou Fan, Jangwon Lee, Mingze Xu, Krishna Kumar Singh, Yong Jae Lee, David J. Crandall, Michael S. Ryoo
Check 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 semi-Siamese branches with specialized early layers instead of fully shared layers (ablation) | two-stream correspondence network · First-/third-person correspondence | Binary AP: 0.453 -> 0.585 | Improved |
| use semi-Siamese branches instead of fully shared layers (ablation) | spatial correspondence network · First-/third-person correspondence | Binary AP: 0.481 -> 0.528 | Improved |
| use semi-Siamese branches instead of fully shared layers (ablation) | temporal correspondence network · First-/third-person correspondence | Binary AP: 0.337 -> 0.389 | Improved |
| replace contrastive loss with triplet loss (ablation) | two-stream semi-Siamese network · First-/third-person correspondence | Binary AP: 0.585 -> 0.621 | Improved |
| replace contrastive loss with triplet loss (ablation) | spatial semi-Siamese network · First-/third-person correspondence | Binary AP: 0.528 -> 0.549 | Improved |
| replace contrastive loss with triplet loss (ablation) | temporal semi-Siamese network · First-/third-person correspondence | Binary AP: 0.389 -> 0.452 | Improved |
| combine spatial and motion cues in a two-stream network (ablation) | semi-Siamese correspondence network · First-/third-person correspondence | Binary AP: 0.528 and 0.389 -> 0.585 | Improved |
| use spatial rather than temporal features (ablation) | triplet semi-Siamese network · First-/third-person correspondence | Binary AP: 0.456 -> 0.549 | Improved |
| remove all layer sharing between the two branches (ablation) | two-stream correspondence network · First-/third-person correspondence | Binary AP: 0.476 -> 0.585 | Hurt |
| add gaze estimates to guide the spatial network (ablation) | spatial correspondence network · First-/third-person correspondence | matching ability | Hurt |
| use triplet loss rather than contrastive loss (ablation) | temporal semi-Siamese network · Multiple wearable cameras | Binary AP: 0.412 -> 0.386 | Hurt |
Closest papers by meaning
- Joint Person Segmentation and Identification in Synchronized First- and Third-person Videos (ECCV 2018 · similarity 0.89)
- Fusing Personal and Environmental Cues for Identification and Segmentation of First-Person Camera Wearers in Third-Person Views (CVPR 2024 · similarity 0.88)
- First Person Action Recognition Using Deep Learned Descriptors (CVPR 2016 · similarity 0.88)
- Actor and Observer: Joint Modeling of First and Third-Person Videos (CVPR 2018 · similarity 0.88)
- Retrieval-Augmented Egocentric Video Captioning (CVPR 2024 · similarity 0.85)
- EgoSim: An Egocentric Multi-view Simulator and Real Dataset for Body-worn Cameras during Motion and Activity (NeurIPS 2024 · similarity 0.84)
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.
