2021
DOI: 10.1007/978-3-030-68787-8_28
|View full text |Cite
|
Sign up to set email alerts
|

Automatic MEP Component Detection with Deep Learning

Abstract: Scan-to-BIM systems convert image and point cloud data into accurate 3D models of buildings. Research on Scan-to-BIM has largely focused on the automated identification of structural components. However, design and maintenance projects require information on a range of other assets including mechanical, electrical, and plumbing (MEP) components. This paper presents a deep learning solution that locates and labels MEP components in 360 • images and phone images, specifically sockets, switches and radiators. The… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1

Citation Types

0
3
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
4
2

Relationship

0
6

Authors

Journals

citations
Cited by 6 publications
(4 citation statements)
references
References 36 publications
(34 reference statements)
0
3
0
Order By: Relevance
“…They independently learn complex features, offer a deeper semantic understanding, and efficiently handle large-scale point cloud datasets (Bapat et al, 2023). Some studies focus on nonstructural components (e.g., mechanical and electrical components), and recent reconstruction efforts in deep learning approaches are starting to include indoor object elements (Kufuor et al, 2021). The reference of (Park et al, 2022) presented a technique for indoor object-based point-net deep learning-based 3D model reconstruction.…”
Section: Deep Learning-based Approachesmentioning
confidence: 99%
“…They independently learn complex features, offer a deeper semantic understanding, and efficiently handle large-scale point cloud datasets (Bapat et al, 2023). Some studies focus on nonstructural components (e.g., mechanical and electrical components), and recent reconstruction efforts in deep learning approaches are starting to include indoor object elements (Kufuor et al, 2021). The reference of (Park et al, 2022) presented a technique for indoor object-based point-net deep learning-based 3D model reconstruction.…”
Section: Deep Learning-based Approachesmentioning
confidence: 99%
“…Yeum et al [61] applied AlexNet [62] to detect the welded joints of a highway sign truss structure. Liang [63] used the faster R-CNN (regions with CNN features) [64] to detect bridge components, while Kufuor et al [65] used it to detect MEP components, including sockets, switches, and radiators, by training both RGB 360 and standard images. Overall, these methods based on R-CNN can achieve high detection accuracy (over 90%) because of the greater availability of labelled data for training.…”
Section: Deep Learning For Shape Detectionmentioning
confidence: 99%
“…Deep learning methods can be generally applied to any object classes, given that there are enough labelled data to train neural networks. Many deep learning methods have been tested in bounding box regression or the PCD semantic and instance segmentation of walls, slabs, doors, windows, columns, beams, or pipes in indoor PCDs, see, e.g., [47,48,65]. These methods can also be applied to the remaining classes of objects; however, there is no scientific evidence of their success.…”
mentioning
confidence: 99%
“…The Faster R-CNN (Girshick, 2015) uses the convolutional network to directly generate candidate regions. It has been applied to detect building electrical instances by training both RGB 360° and standard images (Kufuor et al, 2021). However, this method can only locate the instance position with bounding box.…”
Section: Image-based Instance Detectionmentioning
confidence: 99%