Premium
Pruning deep convolutional neural networks for efficient edge computing in condition assessment of infrastructures
Author(s) -
Wu RihTeng,
Singla Ankush,
Jahanshahi Mohammad R.,
Bertino Elisa,
Ko Bong Jun,
Verma Dinesh
Publication year - 2019
Publication title -
computer‐aided civil and infrastructure engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.773
H-Index - 82
eISSN - 1467-8667
pISSN - 1093-9687
DOI - 10.1111/mice.12449
Subject(s) - computer science , edge computing , convolutional neural network , pruning , enhanced data rates for gsm evolution , context (archaeology) , deep learning , artificial intelligence , edge device , component (thermodynamics) , robot , inference , key (lock) , internet of things , machine learning , distributed computing , computer engineering , embedded system , computer security , operating system , cloud computing , paleontology , physics , thermodynamics , agronomy , biology
Health monitoring of civil infrastructures is a key application of Internet of things (IoT), while edge computing is an important component of IoT. In this context, swarms of autonomous inspection robots, which can replace current manual inspections, are examples of edge devices. Incorporation of pretrained deep learning algorithms into these robots for autonomous damage detection is a challenging problem since these devices are typically limited in computing and memory resources. This study introduces a solution based on network pruning using Taylor expansion to utilize pretrained deep convolutional neural networks for efficient edge computing and incorporation into inspection robots. Results from comprehensive experiments on two pretrained networks (i.e., VGG16 and ResNet18) and two types of prevalent surface defects (i.e., crack and corrosion) are presented and discussed in detail with respect to performance, memory demands, and the inference time for damage detection. It is shown that the proposed approach significantly enhances resource efficiency without decreasing damage detection performance.