Research map / Label-efficient and robust learning / Recognition and long-tail
Weakly Supervised Deep Image Hashing Through Tag Embeddings
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 word2vec tag embeddings rather than binary tag vectors (ablation) | WDHT · NUS-WIDE | mAP: 0.6202 -> 0.6709; 0.627 -> 0.6805; 0.6247 -> 0.6955; 0.6249 -> 0.676 | Helped |
| use word2vec tag embeddings rather than binary tag vectors (ablation) | WDHT · MIR-FLICKR25K | mAP: 0.6365 -> 0.7346; 0.6326 -> 0.743; 0.6373 -> 0.7034; 0.6352 -> 0.7054 | Helped |
| use mean tag-vector aggregation rather than itf or tf (ablation) | WDHT · NUS-WIDE | mAP: 12 bits: 0.6124 / 0.6394 / 0.6709; 24 bits: 0.6323 / 0.6836 / 0.6805; 32 bits: 0.6531 / 0.6881 / 0.6955; 48 bits: 0.6644 | Mixed |
| use the WDHT tag-embedding approach instead of the strongest conventional baseline | WDHT · NUS-WIDE, top 5,000 retrieved | mAP: 0.6551 -> 0.6709; 0.6459 -> 0.6805; 0.6274 -> 0.6955; 0.6478 -> 0.676 | Helped |
| use the WDHT tag-embedding approach instead of the strongest conventional baseline | WDHT · MIR-FLICKR25K, top 5,000 retrieved | mAP: 0.6955 -> 0.7346; 0.7071 -> 0.743; 0.6998 -> 0.7034; 0.6996 -> 0.7054 | Helped |
| increase the hinge-loss weight relative to the quantization-loss weight (ablation) | WDHT · NUS-WIDE validation | mAP | Helped |
| increase the hinge-loss weight relative to the quantization-loss weight (ablation) | WDHT · MIR-FLICKR25K validation | mAP | Helped |
Closest papers by meaning
- Discrete Image Hashing Using Large Weakly Annotated Photo Collections (AAAI 2016 · similarity 0.91)
- Supervised Deep Hashing for Hierarchical Labeled Data (AAAI 2018 · similarity 0.90)
- Weakly Supervised Deep Hyperspherical Quantization for Image Retrieval (AAAI 2021 · similarity 0.88)
- Generalized Debiased Semi-Supervised Hashing for Large-Scale Image Retrieval (AAAI 2025 · similarity 0.86)
- Deep Graph Online Hashing for Multi-Label Image Retrieval (AAAI 2025 · similarity 0.86)
- Deep Unsupervised Hashing via External Guidance (ICML 2025 · similarity 0.85)
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.