MATIS: Masked-Attention Transformers for Surgical Instrument Segmentation

Nicol\'as AyobiAlejandra P\'erez-Rond\'onSantiago Rodr\'iguezPablo Arbel\'aez

Mar 2023

摘要

We propose Masked-Attention Transformers for Surgical Instrument Segmentation(MATIS), a two-stage, fully transformer-based method that leverages modernpixel-wise attention mechanisms for instrument segmentation. MATIS exploits theinstance-level nature of the task by employing a masked attention module thatgenerates and classifies a set of fine instrument region proposals. Our methodincorporates long-term video-level information through video transformers toimprove temporal consistency and enhance mask classification. We validate ourapproach in the two standard public benchmarks, Endovis 2017 and Endovis 2018.Our experiments demonstrate that MATIS' per-frame baseline outperforms previousstate-of-the-art methods and that including our temporal consistency moduleboosts our model's performance further.

展开全部

图表提取

论文十问

参考文献

被引用

提问与回答

暂无人提供速读十问回答

被引用

笔记

问答

领域