Research map / Detection and segmentation / Open-set and few-shot detection
Where Is My Mirror?
Xin Yang, Haiyang Mei, Ke Xu, Xiaopeng Wei, Baocai Yin, Rynson W.H. Lau
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 |
|---|---|---|---|
| use lovasz-hinge loss instead of BCE (ablation) | basic network · MSD test set | IoU: 74.03 -> 75.36 | Improved |
| use lovasz-hinge loss instead of BCE (ablation) | basic network · MSD test set | BER: 10.60 -> 10.44 | Improved |
| use lovasz-hinge loss instead of BCE (ablation) | basic network · MSD test set | pixel accuracy: 0.821 -> 0.820 | No clear effect |
| add multi-scale dilated convolutions (CCFE w/o contrasts) (ablation) | basic + lovasz-hinge · MSD test set | IoU: 75.36 -> 78.59 | Improved |
| add contrast computation to CCFE (full 4 blocks x 4 scales) (ablation) | basic + CCFE w/o contrasts · MSD test set | BER: 8.55 -> 6.39 | Improved |
| add contrast computation to CCFE (full 4 blocks x 4 scales) (ablation) | basic + CCFE w/o contrasts · MSD test set | pixel accuracy: 0.851 -> 0.933 | Improved |
| use 1 CCFE block with 4 scales instead of dilated convs without contrast (ablation) | basic + CCFE · MSD test set | IoU / pixel accuracy: 78.59 -> 76.36 IoU; 0.851 -> 0.882 Acc | Mixed |
| use 4 CCFE blocks with 1 scale of contrast (ablation) | basic + CCFE · MSD test set | IoU / BER: 78.53 IoU; 9.08 BER | Mixed |
| use 4 blocks and 4 scales together instead of 1 block 4 scales (ablation) | MirrorNet · MSD test set | IoU: 76.36 -> 78.95 | Improved |
| add CRF post-processing | MirrorNet · MSD test set | MAE: 0.085 -> 0.065 | Improved |
| add CRF post-processing | MirrorNet · MSD test set | IoU: 78.46 -> 78.95 | Improved |
| add CRF post-processing | BDRAR · MSD test set | IoU: 67.01 -> 67.43 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PiCANet — PiCANet: Learning Pixel-Wise Contextual Attention for Saliency Detection (CVPR 2018) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ZOOM: Learning Video Mirror Detection with Extremely-Weak Supervision (AAAI 2024 · similarity 0.88)
- Detect Any Mirrors: Boosting Learning Reliability on Large-Scale Unlabeled Data with an Iterative Data Engine (CVPR 2025 · similarity 0.87)
- Weakly-Supervised Mirror Detection via Scribble Annotations (AAAI 2024 · similarity 0.86)
- Effective Video Mirror Detection with Inconsistent Motion Cues (CVPR 2024 · similarity 0.85)
- MirrorVerse: Pushing Diffusion Models to Realistically Reflect the World (CVPR 2025 · similarity 0.83)
- MIRROR: Make Your Object-Level Multi-View Generation More Consistent with Training-Free Rectification (ICML 2025 · similarity 0.83)
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.
