Novel low memory footprint DNN models for edge classification of surgeons’ postures

Skill assessment is fundamental to enhance current laparoscopic surgical training and reduce the incidence of musculoskeletal injuries from performing these procedures. Recently, deep neural networks (DNNs) have been used to improve human posture and surgeons’ skills training. While they work well i...

Full description

Saved in:

Bibliographic Details
Main Authors:	Alex Hanneman, Terry Fawden, Marco Branciforte, Maria Celvisia Virzi, Esther L Moss, Luciano Ost, Massimiliano Zecca
Format:	Default Article
Published:	2022
Subjects:	Deep Neural Network Resource-constrained devices Laparoscopic surgeons’ posture
Online Access:	https://hdl.handle.net/2134/20415477.v1
Tags:	Add Tag No Tags, Be the first to tag this record!

id	rr-article-20415477
record_format	Figshare
spelling	rr-article-204154772022-07-13T00:00:00Z Novel low memory footprint DNN models for edge classification of surgeons’ postures Alex Hanneman (13045176) Terry Fawden (13045179) Marco Branciforte (13045182) Maria Celvisia Virzi (13045185) Esther L Moss (7041371) Luciano Ost (4910230) Massimiliano Zecca (1256181) Deep Neural Network Resource-constrained devices Laparoscopic surgeons’ posture <p>Skill assessment is fundamental to enhance current laparoscopic surgical training and reduce the incidence of musculoskeletal injuries from performing these procedures. Recently, deep neural networks (DNNs) have been used to improve human posture and surgeons’ skills training. While they work well in lab, they normally require significant computational power which makes it impossible to use them on edge devices. This paper presents two low memory footprint DNN models used for classifying laparoscopic surgical skill levels at the edge. Trained models were deployed on three Arm Cortex-M processors using the X-Cube-AI and TensorFlow Lite Micro (TFLM) libraries. Results show that the CUBE-AI-based models give the best relative performance, memory footprint, and accuracy trade-offs when executed on the Cortex-M7. </p> 2022-07-13T00:00:00Z Text Journal contribution 2134/20415477.v1 https://figshare.com/articles/journal_contribution/Novel_low_memory_footprint_DNN_models_for_edge_classification_of_surgeons_postures/20415477 All Rights Reserved
institution	Loughborough University
collection	Figshare
topic	Deep Neural Network Resource-constrained devices Laparoscopic surgeons’ posture
spellingShingle	Deep Neural Network Resource-constrained devices Laparoscopic surgeons’ posture Alex Hanneman Terry Fawden Marco Branciforte Maria Celvisia Virzi Esther L Moss Luciano Ost Massimiliano Zecca Novel low memory footprint DNN models for edge classification of surgeons’ postures
description	Skill assessment is fundamental to enhance current laparoscopic surgical training and reduce the incidence of musculoskeletal injuries from performing these procedures. Recently, deep neural networks (DNNs) have been used to improve human posture and surgeons’ skills training. While they work well in lab, they normally require significant computational power which makes it impossible to use them on edge devices. This paper presents two low memory footprint DNN models used for classifying laparoscopic surgical skill levels at the edge. Trained models were deployed on three Arm Cortex-M processors using the X-Cube-AI and TensorFlow Lite Micro (TFLM) libraries. Results show that the CUBE-AI-based models give the best relative performance, memory footprint, and accuracy trade-offs when executed on the Cortex-M7.
format	Default Article
author	Alex Hanneman Terry Fawden Marco Branciforte Maria Celvisia Virzi Esther L Moss Luciano Ost Massimiliano Zecca
author_facet	Alex Hanneman Terry Fawden Marco Branciforte Maria Celvisia Virzi Esther L Moss Luciano Ost Massimiliano Zecca
author_sort	Alex Hanneman (13045176)
title	Novel low memory footprint DNN models for edge classification of surgeons’ postures
title_short	Novel low memory footprint DNN models for edge classification of surgeons’ postures
title_full	Novel low memory footprint DNN models for edge classification of surgeons’ postures
title_fullStr	Novel low memory footprint DNN models for edge classification of surgeons’ postures
title_full_unstemmed	Novel low memory footprint DNN models for edge classification of surgeons’ postures
title_sort	novel low memory footprint dnn models for edge classification of surgeons’ postures
publishDate	2022
url	https://hdl.handle.net/2134/20415477.v1
_version_	1797820017142661120

Novel low memory footprint DNN models for edge classification of surgeons’ postures

Similar Items