Research map / Image generation / Autoregressive and latent generation
MetaScope: Optics-Driven Neural Network for Ultra-Micro Metalens Endoscopy
Wuyang Li, Wentao Pan, Xiaoyuan Liu, Zhendong Luo, Chenxin Li, Hengyu Liu, Din Ping Tsai, Mu Ku Chen, Yixuan Yuan
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 an auxiliary restoration branch guiding segmentation (ablation) | MetaScope · Meta-CVC-Clinic | mDICE: 0.8272 -> 0.8573 | Improved |
| add gradient-guided distillation from DINOv2 (instead of unguided distillation) (ablation) | MetaScope · Meta-CVC-Clinic | mDICE: 0.8750 -> 0.8927 | Improved |
| add Optics-informed Chromatic Correction (OCC) (ablation) | MetaScope · Meta-CVC-Clinic | PSNR: 32.8153 -> 34.0011 | Improved |
| use multi-expert Gaussian (MeG) in OCC (ablation) | MetaScope OCC · Meta-CVC-Clinic | PSNR: 32.8078 -> 34.0011 | Improved |
| add Optics-informed Intensity Adjustment (OIA) with optical prior (ablation) | MetaScope · Meta-CVC-Clinic | PSNR: 34.1341 -> 34.8483 | Improved |
| use simulated spatial plus channel optical priors instead of directly learned embedding (ablation) | MetaScope OIA · Meta-EndoVis-17 | mIoU: 0.8700 -> 0.8802 | Improved |
| use 3 Gaussian experts instead of 1 (ablation) | MetaScope OCC · Meta-EndoVis-17 | mIoU: 0.8682 -> 0.8802 | Improved |
| increase experts from 3 to 5 (ablation) | MetaScope OCC · Meta-EndoVis-17 | mIoU: 0.8802 -> 0.8793 | No clear effect |
| raise restoration loss weight lambda from 0.1 to 1.0 (ablation) | MetaScope · Meta-EndoVis-17 | mIoU: 0.8802 -> 0.8497 | Mixed |
| train one unified model on all five datasets versus DRMI | MetaScope · Meta-CVC-Clinic restoration | PSNR: 34.05 -> 36.47 | Improved |
| fine-tune on the new metalens (5 mm) after zero-shot transfer | MetaScope · Cross-metalens CVC-Clinic | mIoU: 82.82 -> 86.37 | 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.
- Restormer — Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022) · better on 1 of 1 shared result
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
- Leveraging Near-Field Lighting for Monocular Depth Estimation from Endoscopy Videos (ECCV 2024 · similarity 0.84)
- Depth Any Endoscopy: Towards Self-Supervised Generalizable Depth Estimation in Monocular Endoscopy (CVPR 2026 · similarity 0.84)
- Multi-view Masked Contrastive Representation Learning for Endoscopic Video Analysis (NeurIPS 2024 · similarity 0.83)
- Unsupervised Photometric-Consistent Depth Estimation from Endoscopic Monocular Video (AAAI 2025 · similarity 0.82)
- OpticalNet: An Optical Imaging Dataset and Benchmark Beyond the Diffraction Limit (CVPR 2025 · similarity 0.82)
- EndoBench: A Comprehensive Evaluation of Multi-Modal Large Language Models for Endoscopy Analysis (NeurIPS 2025 · similarity 0.81)
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.
