Research map / Label-efficient and robust learning / Recognition and long-tail
Discriminative Invariant Kernel Features: A Bells-and-Whistles-Free Approach to Unsupervised Face Recognition and Pose Estimation
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 |
|---|---|---|---|
| Use discriminative kernel features instead of normalized dot-product pooling | Pose-invariant face matching · Semi-synthetic mugshot, 153 poses | VR: 43.34 -> 74.83 | Helped |
| Replace max pooling with mean pooling (ablation) | DIKF · Semi-synthetic mugshot, 153 poses | VR: 74.83 -> 61.79 | Hurt |
| Replace kernel features with linear discriminative features (ablation) | DIKF · Semi-synthetic mugshot, 153 poses | VR: 74.83 -> 47.23 | Hurt |
| Replace max pooling with mean pooling (ablation) | DIKF · MPIE, pose-invariant identification | Rank-1 identification rate: 75.75 -> 49.5 | Mixed |
| Replace max pooling with mean pooling (ablation) | DIKF · Pose estimation, 15 poses | Rank-1: 86.47 -> 79.07 | Hurt |
| Add a second sequential invariance level for translation, scale, and in-plane rotation (ablation) | DIKF face matching · Semi-synthetic mugshots with additional transformations | verification rate | Helped |
| Model all added transformations together rather than individually (ablation) | Two-level DIKF · Semi-synthetic mugshots with all transformations | verification rate | Hurt |
| Train pose-invariant features using pose-varied templates | DIKF · LFW, Unsupervised Protocol | AUC: 0.9154 | Helped |
Closest papers by meaning
- Towards Pose Invariant Face Recognition in the Wild (CVPR 2018 · similarity 0.88)
- Max-Margin Invariant Features from Transformed Unlabelled Data (NeurIPS 2017 · similarity 0.88)
- Pose-Invariant Face Recognition via Adaptive Angular Distillation (AAAI 2022 · similarity 0.86)
- Pose-Aware Face Recognition in the Wild (CVPR 2016 · similarity 0.86)
- Vec2Face: Scaling Face Dataset Generation with Loosely Constrained Vectors (ICLR 2025 · similarity 0.84)
- Dual Pose-invariant Embeddings: Learning Category and Object-specific Discriminative Representations for Recognition and Retrieval (CVPR 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.