Research map / Label-efficient and robust learning / Recognition and long-tail
Generalising Fine-Grained Sketch-Based Image Retrieval
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 query-conditioned visual-trait descriptors to adapt sketch-photo feature extraction | FG-SBIR on Sketchy · Sketchy | Acc.@ 1: 16.0% -> 22.6% | Helped |
| Use the Sketchy-trained visual-trait dictionary for transfer to shoe sub-categories | FG-SBIR on QMUL-Shoe-V2 · QMUL-Shoe-V2 sub-category transfers | Acc.@ 1: 9.25% -> 13.2%; 14.8% -> 22.2%; 10.6% -> 14.7%; 12.2% -> 15.1%; 7.25% -> 12.2% | Helped |
| Fine-tune the Sketchy visual-trait dictionary on the training sub-category | FG-SBIR on QMUL-Shoe-V2 · QMUL-Shoe-V2 sub-category transfers | Acc.@ 1: 13.2% -> 15.4%; 22.2% -> 24.2%; 14.7% -> 18.4%; 15.1% -> 17.3%; 12.2% -> 15.5% | Mixed |
| Use soft rather than hard descriptor assignment (ablation) | Ours-Full on Sketchy · Sketchy | Acc.@ 1: 20.1% -> 22.6% | Helped |
| Train the visual-trait descriptor on free-hand sketches instead of photo edgemaps (ablation) | Ours-Full on Sketchy · Sketchy | Acc.@ 1: 16.8% -> 22.6% | Helped |
| Use the learned visual-trait descriptor instead of a conventional VAE latent representation (ablation) | FG-SBIR on Sketchy · Sketchy | Acc.@ 1: 12.7% -> 22.6% | Helped |
| Use the learned visual-trait descriptor instead of K-means clusters in VAE space (ablation) | FG-SBIR on Sketchy · Sketchy | Acc.@ 1: 17.6% -> 22.6% | Helped |
| Use visual-trait descriptors instead of category-name word vectors (ablation) | FG-SBIR on Sketchy · Sketchy | Acc.@ 1: 18.0% -> 22.6% | Helped |
| Use visual-trait descriptors instead of sketch-classifier features (ablation) | FG-SBIR on Sketchy · Sketchy | Acc.@ 1: 16.2% -> 22.6% | Helped |
| Increase the descriptor dictionary size from 20 to 300 with hard assignment (ablation) | FG-SBIR on Sketchy · Sketchy | Acc.@ 1: 18.4% -> 20.1% | Helped |
| Increase the descriptor dictionary size from 100 to 300 with soft assignment (ablation) | FG-SBIR on Sketchy · Sketchy | Acc.@ 1: 20.7% -> 22.6% | Helped |
Closest papers by meaning
- Sketch Me That Shoe (CVPR 2016 · similarity 0.94)
- Exploiting Unlabelled Photos for Stronger Fine-Grained SBIR (CVPR 2023 · similarity 0.93)
- ARNet: Self-Supervised FG-SBIR with Unified Sample Feature Alignment and Multi-Scale Token Recycling (AAAI 2025 · similarity 0.92)
- You'll Never Walk Alone: A Sketch and Text Duet for Fine-Grained Image Retrieval (CVPR 2024 · similarity 0.91)
- Asymmetric Mutual Alignment for Unsupervised Zero-Shot Sketch-Based Image Retrieval (AAAI 2024 · similarity 0.90)
- How to Handle Sketch-Abstraction in Sketch-Based Image Retrieval? (CVPR 2024 · similarity 0.89)
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.