Details
Originalsprache | Englisch |
---|---|
Titel des Sammelwerks | Data Augmentation, Labelling, and Imperfections |
Untertitel | Third MICCAI Workshop, DALI 2023, Held in Conjunction with MICCAI 2023, Vancouver, BC, Canada, October 12, 2023, Proceedings |
Herausgeber/-innen | Yuan Xue, Chen Chen, Chao Chen, Lianrui Zuo, Yihao Liu |
Seiten | 95-105 |
Seitenumfang | 11 |
ISBN (elektronisch) | 978-3-031-58171-7 |
Publikationsstatus | Veröffentlicht - 27 Apr. 2024 |
Publikationsreihe
Name | Lecture Notes in Computer Science |
---|---|
Band | 14379 |
ISSN (Print) | 0302-9743 |
ISSN (elektronisch) | 1611-3349 |
Abstract
robotic scrub nurse. For this, image-based deep learning techniques are
effective but usually demand large amounts of annotated data, whose creation
is expensive and time-consuming. In this work, we propose a strategy
based on the copy-paste technique for the generation of reliable synthetic
image training data with a minimal amount of annotation effort.
Our approach enables the efficient in situ creation of datasets for specific
surgeries and contexts. We study the amount of employed manually annotated
data and training set sizes on our model’s performance, as well
as different blending techniques for improved training data. We achieve
91.9 box mAP and 91.6 mask mAP, training solely on synthetic data, in a
real-world scenario. Our evaluation relies on an annotated image dataset
of the wisdom teeth extraction surgery set, created in an actual operating
room. This dataset, the corresponding code, and further data are made
publicly available (https://github.com/Jorebs/Modular-Label-Efficient-
Dataset-Generation-for-Instrument-Detection-for-Robotic-Scrub-Nurses).
ASJC Scopus Sachgebiete
- Mathematik (insg.)
- Theoretische Informatik
- Informatik (insg.)
- Allgemeine Computerwissenschaft
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
Data Augmentation, Labelling, and Imperfections : Third MICCAI Workshop, DALI 2023, Held in Conjunction with MICCAI 2023, Vancouver, BC, Canada, October 12, 2023, Proceedings. Hrsg. / Yuan Xue; Chen Chen; Chao Chen; Lianrui Zuo; Yihao Liu. 2024. S. 95-105 ( Lecture Notes in Computer Science ; Band 14379).
Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung › Peer-Review
}
TY - GEN
T1 - Modular, Label-Efficient Dataset Generation for Instrument Detection for Robotic Scrub Nurses
AU - Badilla Solórzano, Jorge Adrián
AU - Gellrich, Nils Claudius
AU - Seel, Thomas
AU - Ihler, Sontje
PY - 2024/4/27
Y1 - 2024/4/27
N2 - Surgical instrument detection is a fundamental task of arobotic scrub nurse. For this, image-based deep learning techniques areeffective but usually demand large amounts of annotated data, whose creationis expensive and time-consuming. In this work, we propose a strategybased on the copy-paste technique for the generation of reliable syntheticimage training data with a minimal amount of annotation effort.Our approach enables the efficient in situ creation of datasets for specificsurgeries and contexts. We study the amount of employed manually annotateddata and training set sizes on our model’s performance, as wellas different blending techniques for improved training data. We achieve91.9 box mAP and 91.6 mask mAP, training solely on synthetic data, in areal-world scenario. Our evaluation relies on an annotated image datasetof the wisdom teeth extraction surgery set, created in an actual operatingroom. This dataset, the corresponding code, and further data are madepublicly available (https://github.com/Jorebs/Modular-Label-Efficient-Dataset-Generation-for-Instrument-Detection-for-Robotic-Scrub-Nurses).
AB - Surgical instrument detection is a fundamental task of arobotic scrub nurse. For this, image-based deep learning techniques areeffective but usually demand large amounts of annotated data, whose creationis expensive and time-consuming. In this work, we propose a strategybased on the copy-paste technique for the generation of reliable syntheticimage training data with a minimal amount of annotation effort.Our approach enables the efficient in situ creation of datasets for specificsurgeries and contexts. We study the amount of employed manually annotateddata and training set sizes on our model’s performance, as wellas different blending techniques for improved training data. We achieve91.9 box mAP and 91.6 mask mAP, training solely on synthetic data, in areal-world scenario. Our evaluation relies on an annotated image datasetof the wisdom teeth extraction surgery set, created in an actual operatingroom. This dataset, the corresponding code, and further data are madepublicly available (https://github.com/Jorebs/Modular-Label-Efficient-Dataset-Generation-for-Instrument-Detection-for-Robotic-Scrub-Nurses).
KW - Synthetic data
KW - Efficient annotation
KW - Robotic scrub nurse
KW - MBOI
KW - Copy-paste
KW - Deep learning
KW - Robotic Scrub Nurse
UR - http://www.scopus.com/inward/record.url?scp=85192879852&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-58171-7_10
DO - 10.1007/978-3-031-58171-7_10
M3 - Conference contribution
SN - 978-3-031-58170-0
T3 - Lecture Notes in Computer Science
SP - 95
EP - 105
BT - Data Augmentation, Labelling, and Imperfections
A2 - Xue, Yuan
A2 - Chen, Chen
A2 - Chen, Chao
A2 - Zuo, Lianrui
A2 - Liu, Yihao
ER -