Research map / Probabilistic and scientific modeling / Treatment effects
COIN-Matting: Confounder Intervention for Image Matting
Zhaohe Liao, Jiangtong Li, Jun Lan, Huijia Zhu, Weiqiang Wang, Li Niu, Liqing Zhang
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 confounder-intervention (COIN) framework to MatteFormer | MatteFormer · Composition-1K | SAD: 23.8 -> 21.9 | Improved |
| add COIN to MatteFormer | MatteFormer · Distinctions-646 | SAD: 21.9 -> 19.3 | Improved |
| add COIN to DIIM | DIIM · Composition-1K | SAD: 37.5 -> 33.8 | Improved |
| add COIN to guidance-free dugMatting (0-sel) | dugMatting (0-sel) · Composition-1K | SAD: 34.1 -> 30.8 | Improved |
| intervene on transparency confounder only (not joint) (ablation) | COIN(MatteFormer) · Distinctions-646 | SAD: 21.9 -> 20.4 (full COIN 19.3) | Improved |
| intervene on contrast confounder only (ablation) | COIN(MatteFormer) · Distinctions-646 | SAD: 21.9 -> 20.2 (full COIN 19.3) | Improved |
| intervene jointly on contrast and transparency instead of one confounder (ablation) | COIN(MatteFormer) · Distinctions-646 | SAD: 20.2 -> 19.3 | Improved |
| add COIN framework (extra parameters and FLOPs) | MatteFormer · Distinctions-646 | parameters and FLOPs: 44.8M -> 49.5M; 459G -> 508G | Mixed |
Closest papers by meaning
- Attention-Guided Hierarchical Structure Aggregation for Image Matting (CVPR 2020 · similarity 0.87)
- DRIP: Unleashing Diffusion Priors for Joint Foreground and Alpha Prediction in Image Matting (NeurIPS 2024 · similarity 0.87)
- Revisiting Context Aggregation for Image Matting (ICML 2024 · similarity 0.86)
- Training Matting Models Without Alpha Labels (AAAI 2025 · similarity 0.86)
- Unifying Automatic and Interactive Matting with Pretrained ViTs (CVPR 2024 · similarity 0.85)
- In-Context Matting (CVPR 2024 · 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.
