Research map / Label-efficient and robust learning / Recognition and long-tail
Attributable Visual Similarity Learning
Borui Zhang, Wenzhao Zheng, Jie Zhou, Jiwen Lu
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 |
|---|---|---|---|
| Apply AVSL to ProxyAnchor | ProxyAnchor · CUB-200-2011 | Recall@1: 69.7 -> 71.9 | Improved |
| Apply AVSL to Margin | Margin · CUB-200-2011 | Recall@1: 65.6 -> 68.8 | Improved |
| Apply AVSL to ProxyAnchor | ProxyAnchor · Cars196 | Recall@1: 87.7 -> 91.5 | Improved |
| Apply AVSL to Margin | Margin · Cars196 | Recall@1: 78.2 -> 81.1 | Improved |
| Apply AVSL to ProxyAnchor | ProxyAnchor · Stanford Online Products | Recall@1: 78.4 -> 79.6 | Improved |
| Apply AVSL to Margin | Margin · Stanford Online Products | Recall@1: 72.4 -> 76.8 | Improved |
| Add multi-layer embedding losses (ablation) | ProxyAnchor · Cars196 | Recall@1: 87.7 -> 89.7 | Improved |
| Add reliability-based node selection to multi-layer losses (ablation) | ProxyAnchor + M · Cars196 | Recall@1: 89.7 -> 89.9 | Improved |
| Add graph relations and hierarchy-consistent AVSL inference (ablation) | ProxyAnchor + M & R · Cars196 | Recall@1: 89.9 -> 91.5 | Improved |
| Concatenate layer embeddings instead of using AVSL inference (ablation) | ProxyAnchor · Cars196 | Recall@1: 90.6 -> 91.5 | Hurt |
| Remove momentum updating for edge construction (ablation) | ProxyAnchor-AVSL · Cars196 | Recall@1: 91.0 -> 91.5 | Hurt |
| Use a larger edge momentum factor, γ=0.95 rather than γ=0.50 (ablation) | ProxyAnchor-AVSL · Cars196 | Recall@1: 91.5 -> 91.6 | Improved |
Closest papers by meaning
- Conditional Similarity Networks (CVPR 2017 · similarity 0.89)
- Learning Similarity Conditions Without Explicit Supervision (ICCV 2019 · similarity 0.88)
- Interpretable Measures of Conceptual Similarity by Complexity-Constrained Descriptive Auto-Encoding (CVPR 2024 · similarity 0.87)
- The Many Senses of Visual Similarity: A Text-Prompted Image Perceptual Metric (NeurIPS 2026 · similarity 0.87)
- Efficient Discovery and Effective Evaluation of Visual Perceptual Similarity: A Benchmark and Beyond (ICCV 2023 · similarity 0.87)
- Asymmetric Visual Semantic Embedding Framework for Efficient Vision-Language Alignment (AAAI 2025 · 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.
