Research map / Graph learning / Graph algorithms and clustering
Local Connectivity-Based Density Estimation for Face Clustering
Read the paperCheck 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 |
|---|---|---|---|
| Combine local connectivity and feature similarity for density estimation and edge selection (ablation) | Face clustering method · MS-Celeb-1M 584K | FP: 93.61 -> 94.64 | Helped |
| Combine local connectivity and feature similarity for density estimation and edge selection (ablation) | Face clustering method · IJB-B F1845 | FB: 83.74 -> 84.80 | Helped |
| Use local connectivity instead of similarity alone for density estimation and edge selection (ablation) | Face clustering method · DeepFashion | FP: 39.69 -> 39.39 | No clear effect |
| Add local connectivity to similarity-based density estimation (ablation) | Face clustering method · IJB-B F1845 | FB: 81.43 -> 82.38 | Helped |
| Add similarity-based edges alongside density-based edges (ablation) | Face clustering method · IJB-B F1845 | FB: 82.38 -> 84.81 | Helped |
| Use PCENet rather than LCENet for pairwise connectivity estimation (ablation) | Face clustering method · MS-Celeb-1M 584K | FB: 90.69 -> 93.36 | Helped |
| Use PCENet rather than LCENet for pairwise connectivity estimation (ablation) | Face clustering method · DeepFashion | FP: 25.91 -> 41.76 | Helped |
| Use neighbor feature matrices without similarity matrices as PCENet inputs (ablation) | PCENet · MS-Celeb-1M 584K | FP: 94.64 -> 83.35 | Hurt |
| Use both intra-class and inter-class similarities as PCENet inputs (ablation) | PCENet · MS-Celeb-1M 584K | FP: 93.43 -> 94.64 | Helped |
| Use the proposed clustering method instead of Chen et al. | Face clustering · MS-Celeb-1M 584K | FP: 93.22 -> 94.64 | Helped |
| Use the proposed clustering method instead of Chen et al. | Face clustering · MS-Celeb-1M 5.21M | FB: 86.06 -> 87.28 | Helped |
| Use the proposed clustering method instead of the strongest listed baseline | Face clustering · IJB-B F1845 | FP: 82.7 -> 90.8 | Helped |
Closest papers by meaning
- Structure-Aware Face Clustering on a Large-Scale Graph With 107 Nodes (CVPR 2021 · similarity 0.89)
- Learning to Cluster Faces via Confidence and Connectivity Estimation (CVPR 2020 · similarity 0.88)
- Linkage Based Face Clustering via Graph Convolution Network (CVPR 2019 · similarity 0.87)
- Density-Aware Feature Embedding for Face Clustering (CVPR 2020 · similarity 0.87)
- Optimal Graph Clustering without Edge Density Signals (NeurIPS 2025 · similarity 0.84)
- Local Graph Clustering with Noisy Labels (ICLR 2024 · similarity 0.82)
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.