Research map / Label-efficient and robust learning / Self-supervised and contrastive
LEMON: A Large Endoscopic MONocular Dataset and Foundation Model for Perception in Surgical Settings
Chengan Che, Chao Wang, Tom Vercauteren, Sophia Tsoka, Luis C. Garcia-Peraza-Herrera
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 |
|---|---|---|---|
| pretrain on LEMON instead of ImageNet-1K (ablation) | ConvNeXt-L · AutoLaparo (linear probing) | F1: 53.0 -> 66.9 | Improved |
| pretrain on LEMON instead of Cholec80 (51h) (ablation) | ConvNeXt-L · AutoLaparo (linear probing) | F1: 46.9 -> 66.9 | Improved |
| add data curation pipeline (ablation) | LEMON pretraining · AutoLaparo (linear probing) | F1: 61.4 -> 65.9 | Improved |
| add data curation pipeline (ablation) | LEMON pretraining · CholecSeg8k (frozen backbone) | mDice: 67.4 -> 68.7 | Improved |
| add augmented distillation instead of vanilla DINO (ablation) | LemonFM · CholecSeg8k (frozen backbone) | mDice: 68.7 -> 71.9 | Improved |
| add augmented distillation instead of vanilla DINO (ablation) | LemonFM · AutoLaparo (linear probing) | F1: 65.9 -> 66.9 | Improved |
| use ConvNeXt-L instead of ViT-L backbone (ablation) | LemonFM · CholecSeg8k (frozen backbone) | mDice: 61.2 -> 71.9 | Improved |
| use ConvNeXt-L instead of ConvNeXt-B backbone (ablation) | LemonFM · AutoLaparo (linear probing) | F1: 64.5 -> 66.9 | Improved |
| fine-tune backbone instead of freezing | LemonFM · CholecSeg8k | mDice: 71.9 -> 81.3 | Improved |
| train on 100% instead of 50% labels | LemonFM · CholecSeg8k | mDice: 78.2 -> 81.3 | Improved |
| use image-based LemonFM instead of video-based VideoMAEv2 | segmentation · CholecSeg8k | mDice: 65.7 -> 81.3 | Improved |
| use LemonFM-Vid instead of Video Swin Transformer | video classification · LEMON procedure type | mAP: 51.4 -> 57.8 | Improved |
Closest papers by meaning
- EndoBench: A Comprehensive Evaluation of Multi-Modal Large Language Models for Endoscopy Analysis (NeurIPS 2025 · similarity 0.87)
- EgoExOR: An Ego-Exo-Centric Operating Room Dataset for Surgical Activity Understanding (NeurIPS 2025 · similarity 0.86)
- CholecTrack20: A Multi-Perspective Tracking Dataset for Surgical Tools (CVPR 2025 · similarity 0.86)
- Redefining the Laparoscopic Spatial Sense: AI-Based Intra- and Postoperative Measurement from Stereoimages (AAAI 2024 · similarity 0.85)
- MM-OR: A Large Multimodal Operating Room Dataset for Semantic Understanding of High-Intensity Surgical Environments (CVPR 2025 · similarity 0.85)
- SurgCoT: Advancing Spatiotemporal Reasoning in Surgical Videos through a Chain-of-Thought Benchmark (CVPR 2026 · similarity 0.84)
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.
