Name
Title | ||
|---|---|---|
Anchor-guided online meta adaptation for fast one-Shot instrument segmentation from robotic surgical videos |
Abstract | ||
|---|---|---|
•Solving the one-shot instrument segmentation for robotic surgical videos, in which only the first frame mask of each video is needed for fast adaptation.•Designing an anchor-guided online adaptation that continuously adapts the model on first frame mask and subsequent pseudo-masks that generated by anchor matching. The motion-insensitive online supervision can well tackle the fast instrument motion existed in robotic surgical videos.•Proposing to meta-learn the optimal model initialization and learning rate for fast online adaptation through a matching-aware optimization process.•Achieving outstanding segmentation results on two practical scenarios: (i) General to Surgical and (ii) Public to In-house, which can demonstrate the effectiveness and applicability of our method. It also shows great potential for other dense tracking tasks such as tool tip tracking. |
Year | DOI | Venue |
|---|---|---|
2021 | 10.1016/j.media.2021.102240 | Medical Image Analysis |
Keywords | DocType | Volume |
Surgical instrument segmentation,Meta-Learning,Online adaptation,Anchor matching,Robotic surgical video | Journal | 74 |
ISSN | Citations | PageRank |
1361-8415 | 1 | 0.41 |
References | Authors | |
0 | 8 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Zixu Zhao | 1 | 5 | 2.49 |
| Yueming Jin | 2 | 28 | 2.68 |
| Junming Chen | 3 | 7 | 2.94 |
| Bo Lu | 4 | 26 | 12.82 |
| Chi-Fai Ng | 5 | 3 | 1.15 |
| Liu YH | 6 | 1540 | 185.05 |
| Qi Dou | 7 | 837 | 57.52 |
| Pheng-Ann Heng | 8 | 3565 | 280.98 |