Have a personal or library account? Click to login

A novel approach to label road defects in video data: semi-automated video analysis

International Journal on Smart Sensing and Intelligent Systems
Volume 13 (2020): Issue 1 (January 2020)
By:
Open Access
|Apr 2020

Figures & Tables

Figure 1:

Overview of the program workflow.
Overview of the program workflow.

Figure 2:

Algorithm to compress the data and extract information about oscillation.
Algorithm to compress the data and extract information about oscillation.

Figure 3:

Algorithm to calculate the score comparison table.
Algorithm to calculate the score comparison table.

Figure 4:

Algorithm to calculate the final score for each window.
Algorithm to calculate the final score for each window.

Figure 5:

Arrangement of speed sections.
Arrangement of speed sections.

Figure 6:

Score table entries for the minimum and maximum standard deviation of z-acceleration.
Score table entries for the minimum and maximum standard deviation of z-acceleration.

Figure 7:

Image annotation using ‘labelImg’.
Image annotation using ‘labelImg’.

Figure 8:

Comparison between our approach, Maeda, Angulo, and manual labeling. The higher number of events per shown image leads to a lower evaluation time and a less tiring workflow. The significantly higher percentage of severe damages in the labeled set is important for a ground truth with a balanced number of events per class.
Comparison between our approach, Maeda, Angulo, and manual labeling. The higher number of events per shown image leads to a lower evaluation time and a less tiring workflow. The significantly higher percentage of severe damages in the labeled set is important for a ground truth with a balanced number of events per class.

Comparison of evaluations with different thresholds_

Threshold0.280.30.35Manual
Total distance (km)73.9573.9573.958.34
Total video time (min)162.88162.88162.8818.62
Evaluation time (min)4030 17 35
Images shown11989501,117
Cracks105320
Patches6325
Potholes14961
Railroad tracks1100
Speedbump4401
Sunken manholes12954
Sum of events47311631
Images containing events (%) 39.5 34.8322.8
Time per true event (s)51586468
Event every × m1,5732,3854,622 269

Comparison between our approach, Maeda, Angulo, and manual labeling_

Proposed methodMaedaAnguluManual (video)
Presented images119163,664180,3451,117 sec
Containing event (%) 39.5 9.425.22.8
Severe damages (%) 29.8 2.652.643.22
Light damages (%)59.667.9Unknown93.5
Other (%)10.629.45Unknown3.28
Event every × m1,5739736269
DOI: https://doi.org/10.21307/ijssis-2020-007 | Journal eISSN: 1178-5608
Journal RSS Feed
Language: English
Page range: 1 - 9
Submitted on: Dec 5, 2019
Published on: Apr 30, 2020
Published by: Professor Subhas Chandra Mukhopadhyay
In partnership with: Paradigm Publishing Services
Publication frequency: 1 issue per year
Keywords:
Pothole detection,
Road defect detection,
Image labeling,
Acceleration data processing,
Sensor fusion,
Intelligent sensor systems
Related subjects:
Engineering,
Introductions and overviews,
Engineering, other

© 2020 Jakob Thumm, Johannes Masino, Martin Knoche, Frank Gauterin, Markus Reischl, published by Professor Subhas Chandra Mukhopadhyay
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.

Previous articleVolume 13 (2020): Issue 1 (January 2020)Next article