Welcome
Digital library of construction informatics
and information technology in civil engineering and construction
 

Works 

Search Results

Facilitated by the SciX project

Hits 1 to 5 of 5

Bella Nguyen and Ioannis Brilakis

Minimising Misclassifications of Over-Height Vehicles Due to Wind

Abstract: Over-height vehicle strikes with low bridges and tunnels are an ongoing problem worldwide. While previous methods have used vision-based systems to address the over-height warning problem, such methods are sensitive to wind. In this paper, we propose a constraint-based approach to minimise the number of over-height vehicle misclassifications due to windy conditions. The dataset includes a total of 102 over-height vehicles recorded at frame rates of 25 and 30 fps. At this frame rate, we analysed sampling rates to determine the sufficient number of positive frames required to provide accurate warnings to drivers. Optical flow and KLT feature-tracker algorithm was used to detect and track feature points of motion. Motion captured within the region of interest was treated as a standard two-class binary linear classification problem with 1 indicating over-height vehicle presence and 0 indicating noise. The algorithm performed with 100% recall, 83.3% precision and false positive rate of 8.3%.

Keywords: Bridge Strike, Tunnel Strike, Over-Height Vehicle, Over-Height Vehicle Detection System, Bridge Strike Prevention

DOI: https://doi.org/10.24928/JC3-2017/0249

Full text: content.pdf (3,818,410 bytes) (available to registered users only)

Series: jc3:2017 (browse)
Cluster:
Class:
Similar papers:
Sound: N/A.


Ioannis K. Brilakis, and Lucio Soibelman

Shape Recognition Of Linear Construction Entities From Construction Site Images

Abstract:

Keywords:

DOI:

Full text: content.pdf (457,300 bytes) (available to registered users only)

Series: w78:2006 (browse)
Cluster:
Class:
Similar papers:
Sound: N/A.


Ioannis K. Brilakis

Remote Wireless Communications For Construction Management

Abstract:

Keywords:

DOI:

Full text: content.pdf (942,750 bytes) (available to registered users only)

Series: w78:2006 (browse)
Cluster:
Class:
Similar papers:
Sound: N/A.


Ruodan Lu and Ioannis Brilakis

Recursive Segmentation for As-Is Bridge Information Modelling

Abstract: Prior studies reported that the time needed to manually convert a point cloud to an as-is geometric model using cutting edge modelling software is ten times greater than the time needed to obtain the point cloud. The laborious nature of manually modelling infrastructure such as bridges is the reason behind the significant cost of modelling which impedes the proliferation of the usage of Bridge Information Models (BrIM) in Bridge Management Systems. Existing commercial solutions can automatically recognize geometric shapes embedded in segmented point cloud data (PCD) and generate the corresponding IFC objects. Researchers have taken further studies and have additionally automated surface reconstruction through generating parametric surface-based primitives in order to automate the segmentation process. However, surface-based segmentation for bridge modelling is an unsolved problem, which is neither straightforward nor consistent, thus hinders the automation of BrIM.This paper presents a top-down PCD detection solution that follows a knowledge-based heuristic approach for BrIM generation that can semi-automatically segment a bridge point cloud recursively. We leverage bridge domain knowledge as strong priors through a histogram-based algorithm to conduct the tasks of segmentation and classification. We implemented this solution and tested on one highway bridge. The experimental results indicated that the detection precision of this solution is 92%.

Keywords: As-Is Brim, Laser Scanning, Point Cloud Data, Recursive Segmentation

DOI: https://doi.org/10.24928/JC3-2017/0020

Full text: content.pdf (1,880,873 bytes) (available to registered users only)

Series: jc3:2017 (browse)
Cluster:
Class:
Similar papers:
Sound: N/A.


Steven Vick and Ioannis Brilakis

Asphalt Road Layer Detection for Construction Progress Monitoring

Abstract: Transportation construction projects consistently underperform, with an estimated $82.6 billion globally in annual cost overruns. Current progress monitoring practices contribute to this poor performance thanks to their manual, subjective, inaccurate, and time-consuming nature. Automating this task could address these shortfalls and improve project performance. One way to accomplish this automation compares 3D Civil Infrastructure Model design surfaces to 3D point cloud reconstructions of the as-built scene. This requires automated detection of the design surfaces in the as-built data. Research in this area has focused on all-or-nothing detection of structural building components using methods that are a poor fit for large, complex, and closely-layered road design surfaces. These approaches ignore the kind of incremental progress detection needed on transportation projects. This paper proposes a method for detecting large road design surfaces in discrete regions (i.e. increments) of as-built point cloud data, contributing a novel model-guided and sparse hierarchical data structure ('layerTree') that addresses the limitations of existing state-of-the-art methods. The authors collected as-built and as-planned data during construction of a small residential road in Cambridge, UK. A total of 640 experiments on this data examined different combinations of layerTree parameters and classification rules, producing a peak accuracy of 86.62%, peak precision of 80.65%, and peak recall of 92.50%. The most balanced combination produced an accuracy of 86.50%, precision of 68.17%, and recall of 60.99%.

Keywords: Construction Progress Monitoring, Transportation, Drones

DOI: https://doi.org/10.24928/JC3-2017/0329

Full text: content.pdf (6,096,889 bytes) (available to registered users only)

Series: jc3:2017 (browse)
Cluster:
Class:
Similar papers:
Sound: N/A.


No more hits.

 

hosted by University of Ljubljana



includes

W78




© itc.scix.net 2003
Home page of this database login Powered by SciX Open Publishing Services 1.002 February 16, 2003