Research map / Label-efficient and robust learning / Noisy labels
Learning Semantic Correspondence with Sparse Annotations
Shuaiyi Huang, Luyu Yang, Bo He, Songyang Zhang, Xuming He, Abhinav Shrivastava
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 Efficient-SCE spatial context encoder (ablation) | base network · SPair-71k | PCK: 14.7 -> 33.4 | Improved |
| add high-resolution loss (ablation) | network with Efficient-SCE · SPair-71k | PCK: 33.4 -> 40.6 | Improved |
| finetune the backbone (ablation) | network with Efficient-SCE and HRLoss · SPair-71k | PCK: 40.6 -> 49.8 | Improved |
| add teacher-student learning without denoising (ablation) | base network · SPair-71k | PCK: 49.8 -> 49.7 | No clear effect |
| add dynamic label selection (ablation) | teacher-student model · SPair-71k | PCK: 49.7 -> 51.5 | Improved |
| add label mask dilation (ablation) | teacher-student with dynamic selection · SPair-71k | PCK: 51.5 -> 55.3 | Improved |
| use single offline teacher (ablation) | base network · SPair-71k | PCK: 49.8 -> 52.4 | Improved |
| use mutual online teachers instead of single offline teacher (ablation) | teacher-student · SPair-71k | PCK: 52.4 -> 55.3 | Improved |
| use dilation kernel size 7 instead of 3 (ablation) | label mask dilation · SPair-71k | PCK: 54.0 -> 55.3 | Improved |
| use dilation kernel size 15 instead of 7 (ablation) | label mask dilation · SPair-71k | PCK: 55.3 -> 52.0 | Hurt |
| use ours (MT) instead of CATs | PCK on SPair-71k · SPair-71k | PCK: 49.9 -> 55.3 | Improved |
| use ours (MT) instead of CATs at alpha 0.15 | PF-WILLOW · PF-WILLOW | PCK@0.15: 90.3 -> 89.8 | Hurt |
Closest papers by meaning
- Joint Inference in Weakly-Annotated Image Datasets via Dense Correspondence (IJCV 2016 · similarity 0.90)
- Do It Yourself: Learning Semantic Correspondence from Pseudo-Labels (ICCV 2025 · similarity 0.89)
- Semi-Supervised Learning of Semantic Correspondence With Pseudo-Labels (CVPR 2022 · similarity 0.89)
- Images as Noisy Labels: Unleashing the Potential of the Diffusion Model for Open-Vocabulary Semantic Segmentation (ICCV 2025 · similarity 0.88)
- Hard Labels In! Rethinking the Role of Hard Labels in Mitigating Local Semantic Drift (ICML 2026 · similarity 0.88)
- Point Cloud Semantic Segmentation with Sparse and Inhomogeneous Annotations (AAAI 2025 · similarity 0.86)
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.
