Research map / Label-efficient and robust learning / Recognition and long-tail
Person Recognition in Personal Photo Collections
Seong Joon Oh, Rodrigo Benenson, Mario Fritz, Bernt Schiele
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 CASIA and CACD face data (ablation) | head cue h · PIPA val | accuracy: 83.88 -> 86.26 | Improved |
| add CACD data only (ablation) | head cue h · PIPA val | accuracy: 83.88 -> 84.88 | Improved |
| add the scene cue to face/head/upper/body (ablation) | P · PIPA val | accuracy: 91.14 -> 91.16 | No clear effect |
| add the body cue to f+h+u (ablation) | f+h+u · PIPA val | accuracy: 90.65 -> 91.14 | Improved |
| use upper body instead of full body (ablation) | body cue · PIPA val | accuracy: 80.81 -> 84.76 | Improved |
| use L2-normalised weighted concatenation of naeil and DeepID2+ instead of naive concatenation (ablation) | naeil · PIPA val (Day split) | accuracy: 51.02 -> 70.58 | Improved |
| L2-normalise before concatenating head and DeepID2+ features (ablation) | h + h_deepid · PIPA val (Day split) | accuracy: 47.86 -> 66.91 | Improved |
| add DeepID2+ face module to naeil (weighted) | naeil · PIPA test (Day split) | accuracy: 46.54 -> 70.58 | Improved |
| add DeepID2+ face module to naeil (weighted) | naeil · PIPA test (Original split) | accuracy: 86.78 -> 90.42 | Improved |
| pre-train scene cue on Places instead of ImageNet (raw features) (ablation) | scene cue · PIPA val | accuracy: 26.54 -> 27.37 | Improved |
| fine-tune PlacesNet on person recognition (ablation) | scene cue · PIPA val | accuracy: 27.37 -> 25.62 | Hurt |
Closest papers by meaning
- A Multi-Level Contextual Model For Person Recognition in Photo Albums (CVPR 2016 · similarity 0.90)
- Pose-Aware Person Recognition (CVPR 2017 · similarity 0.88)
- Unifying Identification and Context Learning for Person Recognition (CVPR 2018 · similarity 0.87)
- Faceless Person Recognition: Privacy Implications in Social Media (ECCV 2016 · similarity 0.87)
- Sequential Person Recognition in Photo Albums With a Recurrent Network (CVPR 2017 · similarity 0.86)
- 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.
