Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Adaptive Margin Diversity Regularizer for handling Data Imbalance in Zero-Shot SBIR
Titir Dutta, Anurag Singh, Soma Biswas
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 AMDReg | SEM-PCYC · TU-Berlin Ext. ZS-SBIR | MAP@all: 0.297 -> 0.330 | Improved |
| add AMDReg | SEM-PCYC · Sketchy Ext. ZS-SBIR | MAP@all: 0.349 -> 0.397 | Improved |
| add AMDReg | Style-guide · TU-Berlin Ext. ZS-SBIR | MAP@all: 0.254 -> 0.291 | Improved |
| add AMDReg | Style-guide · Sketchy Ext. ZS-SBIR | MAP@all: 0.375 -> 0.410 | Improved |
| add AMDReg | SAKE · TU-Berlin Ext. ZS-SBIR | MAP@all: 0.428 -> 0.447 | Improved |
| add AMDReg | SAKE · Sketchy Ext. ZS-SBIR | MAP@all: 0.547 -> 0.551 | No clear effect |
| add AMDReg | SEM-PCYC · TU-Berlin Ext. generalized ZS-SBIR | MAP@all: 0.192 -> 0.245 | Improved |
| add AMDReg | SEM-PCYC · Sketchy Ext. generalized ZS-SBIR | MAP@all: 0.307 -> 0.320 | Improved |
| introduce step imbalance p=100 into training data | Baseline Model [7] · mini-Sketchy ZS-SBIR | MAP@200: 0.395 -> 0.156 | Hurt |
| add AMDReg to baseline under step imbalance p=100 | Baseline Model [7] · mini-Sketchy ZS-SBIR | MAP@200: 0.156 -> 0.218 | Improved |
| add adaptive imbalance margin to the diversity regularizer (AMDReg vs Diversity-Regularizer) (ablation) | Baseline Model [7] · mini-Sketchy ZS-SBIR long-tailed p=10 | MAP@200: 0.296 -> 0.332 | Improved |
| use AMDReg instead of LDAM-CE loss (ablation) | Baseline Model [7] · mini-Sketchy ZS-SBIR long-tailed p=100 | MAP@200: 0.234 -> 0.240 | No clear effect |
Closest papers by meaning
- Asymmetric Mutual Alignment for Unsupervised Zero-Shot Sketch-Based Image Retrieval (AAAI 2024 · similarity 0.89)
- A Zero-Shot Framework for Sketch based Image Retrieval (ECCV 2018 · similarity 0.88)
- CLIP for All Things Zero-Shot Sketch-Based Image Retrieval, Fine-Grained or Not (CVPR 2023 · similarity 0.88)
- Semantic-Aware Knowledge Preservation for Zero-Shot Sketch-Based Image Retrieval (ICCV 2019 · similarity 0.88)
- Doodle to Search: Practical Zero-Shot Sketch-Based Image Retrieval (CVPR 2019 · similarity 0.88)
- ARNet: Self-Supervised FG-SBIR with Unified Sample Feature Alignment and Multi-Scale Token Recycling (AAAI 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.
