Research map / Detection and segmentation / Open-set and few-shot detection
Learning Transparent Object Matting
Guanying Chen, Kai Han, Kwan-Yee K. Wong
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 refractive flow regression loss (ablation) | CoarseNet · Synthetic ablation | F-EPE: 3.9 / 26.5 -> 2.2 / 15.4 | Improved |
| add cross-link between decoders (ablation) | CoarseNet · Synthetic ablation | F-EPE: 2.5 / 17.2 -> 2.2 / 15.4 | Improved |
| add multi-scale loss (ablation) | CoarseNet · Synthetic ablation | A-MSE: 0.69 -> 0.28 | Improved |
| add image reconstruction loss in coarse stage (ablation) | CoarseNet · Synthetic ablation | I-MSE: 0.22 -> 0.18 | Improved |
| add image reconstruction loss in coarse stage (ablation) | CoarseNet · Synthetic ablation | F-EPE: 2.3 / 15.7 -> 2.2 / 15.4 | Improved |
| add RefineNet after CoarseNet (ablation) | CoarseNet · Synthetic ablation | F-EPE: 2.2 / 15.4 -> 2.0 / 13.7 | Improved |
| add image reconstruction loss in refinement stage (ablation) | RefineNet · Synthetic ablation | F-EPE: 2.0 / 13.7 -> 2.0 / 13.9 | No clear effect |
| add RefineNet (full TOM-Net) (ablation) | CoarseNet · Synthetic test average | F-EPE: 2.9 / 20.1 -> 2.7 / 18.6 | Improved |
| add RefineNet (full TOM-Net) | CoarseNet · Real data average | PSNR: 21.91 -> 21.84 | No clear effect |
| add trimap input at test time | TOM-Net · Real data average | PSNR: 21.84 -> 23.53 | Improved |
| add background image input | TOM-Net · Real data average | PSNR: 21.84 -> 24.61 | Improved |
| add background image input | TOM-Net · Synthetic test average | F-EPE: 2.7 / 18.6 -> 2.2 / 16.2 | Improved |
Closest papers by meaning
- TOM-Net: Learning Transparent Object Matting From a Single Image (CVPR 2018 · similarity 0.99)
- TransMatting: Enhancing Transparent Objects Matting with Transformers (ECCV 2022 · similarity 0.91)
- Image Matting without Matting-Specific Annotations via Eikonal Fields (NeurIPS 2026 · similarity 0.89)
- Deep Image Matting (CVPR 2017 · similarity 0.88)
- DRIP: Unleashing Diffusion Priors for Joint Foreground and Alpha Prediction in Image Matting (NeurIPS 2024 · similarity 0.87)
- Training Matting Models Without Alpha Labels (AAAI 2025 · similarity 0.87)
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.
