Research map / Label-efficient and robust learning / Recognition and long-tail
Sequential Person Recognition in Photo Albums With a Recurrent Network
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 relation context by predicting people’s identities sequentially (ablation) | Appearance-only person recognition · PIPA Original, Album, Time, Day | overall accuracy: 75.43 -> 76.73; 68.31 -> 68.85; 57.19 -> 58.57; 36.37 -> 40.39 | Helped |
| add scene context through the global image feature (ablation) | Ours-relation · PIPA Original, Album, Time, Day | overall accuracy: 76.73 -> 81.75; 68.85 -> 74.21; 58.57 -> 63.73; 40.39 -> 42.75 | Helped |
| use element-wise max rather than addition for the joint embedding (ablation) | Sequential person recognition model · PIPA Original, Album, Time, Day | accuracy: 81.51 -> 81.75; 73.21 -> 74.21; 62.97 -> 63.73; 43.15 -> 42.75 | Mixed |
| fuse head and upper-body predictions by averaging probabilities (ablation) | Head-region sequential model · PIPA Original, Album, Time, Day | accuracy: 81.75 -> 84.93; 74.21 -> 78.25; 63.73 -> 66.43; 42.75 -> 43.73 | Helped |
| use max rather than average to fuse regional probabilities (ablation) | Head and upper-body models · PIPA Original, Album, Time, Day | accuracy: 84.93 -> 84.07; 78.25 -> 75.88; 66.43 -> 65.63; 43.73 -> 43.55 | Hurt |
| concatenate head and upper-body features (ablation) | Sequential person recognition model · PIPA Original, Album, Time, Day | accuracy: 84.93 -> 82.86; 78.25 -> 74.66; 66.43 -> 63.62; 43.73 -> 41.56 | Hurt |
| use the head region rather than the upper-body region (ablation) | Sequential person recognition model · PIPA Original, Album, Time, Day | accuracy: 79.92 vs 81.75; 70.78 vs 74.21; 58.80 vs 63.73; 34.61 vs 42.75 | Helped |
| use sequential relation and scene context with head and upper-body average fusion | Oh et al. person recognition system · PIPA Original, Album, Time, Day | accuracy: 83.63 -> 84.93; 74.52 -> 78.25; 63.98 -> 66.43; 38.94 -> 43.73 | Helped |
Closest papers by meaning
- A Multi-Level Contextual Model For Person Recognition in Photo Albums (CVPR 2016 · similarity 0.89)
- Unifying Identification and Context Learning for Person Recognition (CVPR 2018 · similarity 0.88)
- Pose-Aware Person Recognition (CVPR 2017 · similarity 0.86)
- Relation-Aware Global Attention for Person Re-Identification (CVPR 2020 · similarity 0.86)
- Recurrent Attention Models for Depth-Based Person Identification (CVPR 2016 · similarity 0.85)
- Learning from Synchronization: Self-Supervised Uncalibrated Multi-View Person Association in Challenging Scenes (CVPR 2025 · 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.