Research map / Label-efficient and robust learning / Recognition and long-tail
Attribute-Aware Deep Hashing With Self-Consistency for Large-Scale Fine-Grained Image Retrieval
Xiu-Shen Wei, Yang Shen, Xuhao Sun, Peng Wang, Yuxin Peng
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 attention guidance to the vanilla backbone (ablation) | ResNet-50 hashing · CUB200-2011 (48 bits) | mAP (%): 65.43 -> 76.10 | Improved |
| add unsupervised attribute-guided feature reconstruction (ablation) | A2-Net with attention · CUB200-2011 (48 bits) | mAP (%): 76.10 -> 77.10 | Improved |
| add image reconstruction from only the first feature (ablation) | A2-Net with feature reconstruction · CUB200-2011 (48 bits) | mAP (%): 77.10 -> 77.15 | No clear effect |
| add image reconstruction from only the second feature (ablation) | A2-Net with feature reconstruction · CUB200-2011 (48 bits) | mAP (%): 77.10 -> 79.54 | Improved |
| add dual image reconstruction (both features) (ablation) | A2-Net with feature reconstruction · CUB200-2011 (48 bits) | mAP (%): 77.10 -> 81.83 | Improved |
| add feature decorrelation after dual image reconstruction (ablation) | A2-Net++ · CUB200-2011 (48 bits) | mAP (%): 81.83 -> 82.71 | Improved |
| add image reconstruction self-consistency (A2-Net++ over A2-Net) (ablation) | A2-Net · VegFru (32 bits) | mAP (%): 52.75 -> 73.38 | Improved |
| add image reconstruction self-consistency (A2-Net++ over A2-Net) (ablation) | A2-Net · Aircraft (12 bits) | mAP (%): 40.00 -> 57.53 | Improved |
| add attribute-guided reconstruction to baseline on MNIST-CIFAR (ablation) | Baseline hashing · MNIST-CIFAR CIFAR-only (48 bits) | mAP (%): 21.33 -> 23.54 | Improved |
| add image reconstruction on top of A2-Net on MNIST-CIFAR (ablation) | A2-Net · MNIST-CIFAR CIFAR-only (48 bits) | mAP (%): 23.54 -> 29.40 | Improved |
| add reconstruction on MNIST-only test (ablation) | Baseline hashing · MNIST-only (48 bits) | mAP (%): 99.52 -> 99.52 | No clear effect |
| vary eta in {0.05, 0.10, 0.20, 0.30, 0.40} (ablation) | A2-Net++ · CUB200-2011 and Aircraft (48 bits) | mAP (%) | No clear effect |
Closest papers by meaning
- A$^2$-Net: Learning Attribute-Aware Hash Codes for Large-Scale Fine-Grained Image Retrieval (NeurIPS 2021 · similarity 0.96)
- Learning Attribute-Aware Hash Codes for Fine-Grained Image Retrieval via Query Optimization (ICML 2025 · similarity 0.95)
- Learning Multifunctional Binary Codes for Both Category and Attribute Oriented Retrieval Tasks (CVPR 2017 · similarity 0.91)
- Characteristics Matching Based Hash Codes Generation for Efficient Fine-grained Image Retrieval (CVPR 2024 · similarity 0.90)
- An Asymmetric Augmented Self-Supervised Learning Method for Unsupervised Fine-Grained Image Hashing (CVPR 2024 · similarity 0.89)
- Pairwise-Label-Based Deep Incremental Hashing with Simultaneous Code Expansion (AAAI 2024 · similarity 0.87)
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.
