Table 1 Comparison of annotated subsets in the Cataract-1K dataset with existing datasets for semantic segmentation and phase recognition in cataract surgery.
From: Cataract-1K Dataset for Deep-Learning-Assisted Analysis of Cataract Surgery Videos
Dataset | CaDIS20 | CatRel15 | Cataract-1K | |
|---|---|---|---|---|
Acquisition Time | 2015 | 2017–2018 | 2021–2023 | |
Task | Semantic Segmentation | Phase Recognition | Semantic Segmentation | Phase Recognition |
Videos | 25 | 22 | 30 | 56 |
Annotations | 4670 | Frame-wise | 2256 frames | Frame-wise |
Labels | Anatomy, Instruments, Other | Four Relevant Phases vs. Rest | Anatomy, Instruments | Thirteen Phases |
Resolution | 960 × 540 | 224 × 224 | 1024 × 768 | 1024 × 768 |
Frame-Rate | N/A | 25 | N/A | 30 |
