Contribution to a conference proceedings/Contribution to a book

DZNE-2025-01010

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png AutoSNAP: Automatically Learning Neural Architectures for Instrument Pose Estimation

Martel, A. L. (Editor) ; Abolmaesumi, P. (Editor) ; Stoyanov, D. (Editor) ; Mateus, D. (Editor) ; Zuluaga, M. A. (Editor) ; Zhou, S. K. (Editor) ; Racoceanu, D. (Editor) ; Joskowicz, L. (Editor) ; Kügler, D.DZNE* ; Uecker, M. ; Kuijper, A. ; Mukhopadhyay, A.

2020
Springer International Publishing Cham
ISBN: 978-3-030-59715-3 (print), 978-3-030-59716-0 (electronic)

This record in other databases:

Please use a persistent id in citations: doi:10.1007/978-3-030-59716-0_36

Abstract: Despite recent successes, the advances in Deep Learning have not yet been fully translated to Computer Assisted Intervention (CAI) problems such as pose estimation of surgical instruments. Currently, neural architectures for classification and segmentation tasks are adopted ignoring significant discrepancies between CAI and these tasks. We propose an automatic framework (AutoSNAP) for instrument pose estimation problems, which discovers and learns architectures for neural networks. We introduce 1) an efficient testing environment for pose estimation, 2) a powerful architecture representation based on novel Symbolic Neural Architecture Patterns (SNAPs), and 3) an optimization of the architecture using an efficient search scheme. Using AutoSNAP, we discover an improved architecture (SNAPNet) which outperforms both the hand-engineered i3PosNet and the state-of-the-art architecture search method DARTS.

---

Contributing Institute(s):

1. Artificial Intelligence in Medicine (AG Reuter)

Research Program(s):

1. 354 - Disease Prevention and Healthy Aging (POF4-354) (POF4-354)

---

Database coverage:
Nationallizenz ; SCOPUS

---