Research map / Detection and segmentation / Open-set and few-shot detection
A Multi-Task Network for Joint Specular Highlight Detection and Removal
Gang Fu, Qing Zhang, Lei Zhu, Ping Li, Chunxia Xiao
Check 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 highlight detection to the multi-task network (ablation) | JSHDR · SHIQ | PSNR: 32.09 -> 34.13 | Improved |
| Add highlight detection and highlight estimation to the multi-task network (ablation) | JSHDR · SHIQ | PSNR: 31.81 -> 34.13 | Improved |
| Use joint detection, estimation, and removal instead of highlight estimation alone (ablation) | JSHDR · SHIQ | PSNR: 26.19 -> 34.13 | Improved |
| Use four DSCFA blocks instead of one (ablation) | JSHDR · SHIQ | PSNR: 31.53 -> 34.13 | Improved |
| Use four DSCFA blocks instead of two (ablation) | JSHDR · SHIQ | PSNR: 33.67 -> 34.13 | Improved |
| Use four DSCFA blocks instead of three (ablation) | JSHDR · SHIQ | PSNR: 34.02 -> 34.13 | Improved |
| Use DSCFA modules instead of simple convolution operations (ablation) | JSHDR · SHIQ | PSNR: 18.34 -> 34.13 | Improved |
| Handle images with colored lighting using the proposed network | JSHDR · colored-lighting images | visual highlight removal quality | Hurt |
Closest papers by meaning
- Towards High-Quality Specular Highlight Removal by Leveraging Large-Scale Synthetic Data (ICCV 2023 · similarity 0.91)
- One-Step Specular Highlight Removal with Adapted Diffusion Models (ICCV 2025 · similarity 0.90)
- Hierarchical Adaptive Filtering Network for Text Image Specular Highlight Removal (CVPR 2025 · similarity 0.89)
- Neural Solver of Dichromatic Reflection Model for Specular Highlight Removal (ICCV 2025 · similarity 0.89)
- SHDocs: A dataset, benchmark, and method to efficiently generate high-quality, real-world specular highlight data with near-perfect alignment (NeurIPS 2024 · similarity 0.89)
- PHR-DIFF: Portrait Highlights Removal via Patch-aware Diffusion Model (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.
