Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Exploiting a Joint Embedding Space for Generalized Zero-Shot Semantic Segmentation
Donghyeon Baek, Youngmin Oh, Bumsub Ham
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 |
|---|---|---|---|
| replace center regression with BAR loss (ablation) | JoEm NN classifier · PASCAL Context unseen-4 | hIoU: 15.8 -> 16.7 | Improved |
| add SC loss to center regression (ablation) | JoEm NN classifier · PASCAL Context unseen-4 | hIoU: 15.8 -> 17.8 | Improved |
| add SC loss to BAR loss (ablation) | JoEm NN classifier · PASCAL Context unseen-4 | hIoU: 16.7 -> 19.0 | Improved |
| add SC loss to BAR loss (ablation) | JoEm NN classifier · PASCAL Context unseen-4 | mIoU_S: 37.9 -> 36.2 | Mixed |
| add calibrated stacking at inference (ablation) | JoEm (BAR+SC) · PASCAL Context unseen-4 | hIoU: 19.0 -> 32.3 | Improved |
| add Apollonius calibration at inference (ablation) | JoEm (BAR+SC) · PASCAL Context unseen-4 | hIoU: 19.0 -> 33.7 | Improved |
| use Apollonius calibration instead of calibrated stacking (ablation) | JoEm (BAR+SC) · PASCAL Context unseen-4 | mIoU_U: 29.1 -> 31.8 | Improved |
| use Apollonius calibration instead of calibrated stacking (ablation) | JoEm (BAR+SC) · PASCAL Context unseen-4 | mIoU_S: 36.2 -> 35.7 | Hurt |
| train visual and semantic encoders jointly instead of two-stage S->V (ablation) | JoEm · PASCAL VOC unseen-4 | hIoU: 41.5 -> 44.6 | Improved |
| use BAR + SC instead of center loss in two-stage S->V (ablation) | S->V two-stage · PASCAL VOC unseen-4 | hIoU: 31.2 -> 41.5 | Improved |
| train ZS3Net generator on features from JoEm visual encoder (ablation) | ZS3Net · PASCAL VOC unseen-4 | hIoU: 40.6 -> 43.4 | Improved |
| use JoEm instead of CSRL | JoEm · PASCAL VOC unseen-2 | hIoU: 56.3 -> 53.1 | Hurt |
Closest papers by meaning
- Zero-Shot Semantic Segmentation (NeurIPS 2019 · similarity 0.94)
- Improving Generalized Zero-Shot Learning by Exploring the Diverse Semantics from External Class Names (CVPR 2024 · similarity 0.90)
- TPR: Topology-Preserving Reservoirs for Generalized Zero-Shot Learning (NeurIPS 2024 · similarity 0.90)
- ZeroDiff: Solidified Visual-semantic Correlation in Zero-Shot Learning (ICLR 2025 · similarity 0.89)
- Epsilon: Exploring Comprehensive Visual-Semantic Projection for Multi-Label Zero-Shot Learning (AAAI 2025 · similarity 0.89)
- ZeroMamba: Exploring Visual State Space Model for Zero-Shot Learning (AAAI 2025 · 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.
