Research map / Label-efficient and robust learning / Self-supervised and contrastive
DLBD: A Self-Supervised Direct-Learned Binary Descriptor
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 |
|---|---|---|---|
| Use direct binary learning with BTL and UWT-Xent instead of UDBD | DLBD · CIFAR10 image retrieval | mAP@1000 (%): 39.60 -> 58.58 | Helped |
| Use DLBD instead of BLCD | Unsupervised binary descriptors · Phototour patch matching | FPR95 (mean): 8.53 -> 13.18 | Hurt |
| Use DLBD with linear evaluation instead of DeepBit with SVM | Flower17 image classification · Flower17 | Accuracy (%): 89.2 -> 94.0 | Helped |
| Replace sign binarization with BTL | CIFAR10 binary descriptors · CIFAR10 | mAP@1000 (%): 51.62 -> 58.58 | Helped |
| Replace softsign binarization with BTL | CIFAR10 binary descriptors · CIFAR10 | mAP@1000 (%): 55.52 -> 58.58 | Helped |
| Add Eq. (6) gradient normalization to BTL (ablation) | BTL · CIFAR10, 16-bit | mAP@1000 (%): 39.39 -> 58.58 | Helped |
| Add Eq. (6) gradient normalization to BTL (ablation) | BTL · CIFAR10, 16-bit | mAP@1000 (%): 26.06 -> 58.58 | Helped |
| Add BTL to SimCLR (ablation) | SimCLR · CIFAR10, 16-bit | mAP@1000 (%): 60.8 -> 61.6 | No clear effect |
| Add BTL to SimCLR (ablation) | SimCLR · CIFAR10, 32-bit | mAP@1000 (%): 62.3 -> 63.5 | Helped |
| Add BTL to SimCLR (ablation) | SimCLR · CIFAR10, 64-bit | Accuracy (%): 69.2 -> 70.8 | Helped |
| Add UWT-Xent to SimCLR + BTL (ablation) | SimCLR + BTL · CIFAR10, 16-bit | mAP@1000 (%): 61.6 -> 58.6 | Hurt |
| Add UWT-Xent to SimCLR + BTL (ablation) | SimCLR + BTL · CIFAR10, 64-bit | mAP@1000 (%): 63.4 -> 62.9 | Hurt |
Closest papers by meaning
- Learning Compact Binary Descriptors With Unsupervised Deep Neural Networks (CVPR 2016 · similarity 0.87)
- Learning Deep Binary Descriptor With Multi-Quantization (CVPR 2017 · similarity 0.86)
- BinGAN: Learning Compact Binary Descriptors with a Regularized GAN (NeurIPS 2018 · similarity 0.86)
- SDGMNet: Statistic-Based Dynamic Gradient Modulation for Local Descriptor Learning (AAAI 2024 · similarity 0.83)
- Efficient Multitask Dense Predictor via Binarization (CVPR 2024 · similarity 0.83)
- BVT-IMA: Binary Vision Transformer with Information-Modified Attention (AAAI 2024 · 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.