Research map / Label-efficient and robust learning / Recognition and long-tail
L2-Net: Deep Learning of Discriminative Patch Descriptor in Euclidean Space
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 CS L2-Net+ instead of CS SNet-GLoss | Brown patch matching · Brown | false positive rate at 95% recall: 2.71 -> 1.76 | Helped |
| Add data augmentation (ablation) | L2-Net · Brown | false positive rate at 95% recall: 3.23 -> 2.22 | Helped |
| Use the central-surround two-tower structure (ablation) | L2-Net · Brown | false positive rate at 95% recall: 3.23 -> 2.61 | Helped |
| Train on HPatches rather than Brown | L2-Net · Brown | false positive rate at 95% recall | Mixed |
| Remove the descriptor compactness term E2 (ablation) | L2-Net · Brown | convergence | Hurt |
| Add descriptor compactness term E2 (ablation) | L2-Net with progressive sampling · Brown | generalization | Helped |
| Remove discriminative intermediate feature supervision E3 (ablation) | L2-Net · Brown | AUC | Hurt |
| Apply DIF after every batch-normalization layer (ablation) | L2-Net · Brown | AUC | Hurt |
| Learn batch-normalization weighting and bias rather than fixing them (ablation) | L2-Net · Brown | AUC | Hurt |
| Use central-surround cropping (ablation) | L2-Net · Oxford image sequences | mAP | Mixed |
| Use L2-Net on blurred image sequences | L2-Net · Oxford bikes sequence | mAP | Hurt |
Closest papers by meaning
- Explicit Spatial Encoding for Deep Local Descriptors (CVPR 2019 · similarity 0.89)
- Kernelized Subspace Pooling for Deep Local Descriptors (CVPR 2018 · similarity 0.86)
- Local Descriptors Optimized for Average Precision (CVPR 2018 · similarity 0.85)
- Patch Reordering: A NovelWay to Achieve Rotation and Translation Invariance in Convolutional Neural Networks (AAAI 2017 · similarity 0.85)
- Efficient Learning of CNNs using Patch Based Features (ICML 2022 · similarity 0.85)
- Latent Zoning Network: A Unified Principle for Generative Modeling, Representation Learning, and Classification (NeurIPS 2025 · similarity 0.83)
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.