Georeferencing of Road Infrastructure from Photographs using Computer Vision and Deep Learning for Road Safety Applications

Georeferenced information of road infrastructure is crucial for road safety analysis. Unfortunately, for essential structures, such as fences and crash barriers, exact location information and extent is often not available hindering any kind of spatial analysis. For a GIS-based study on wildlife-vehicle collisions (WVCs) and, therein, the impact of these structures, we developed a method to derive this data from video-based road inspections. A deep learning approach was applied to identify fences and barriers in photos and to estimate the extent and location, based on the photos’ metadata and perspective. We used GIS-based analysis and geometric functions to convert this data into georeferenced line segments. For a road network of 113 km, we were able to identify over 88% of all barrier lines. The main problems for the application of this method are infrastructure invisible from the road or hidden behind vegetation, and the small sections along the streets covered by photos not depicting the tops of higher dams or slopes.

[1]  Tõnu Oja,et al.  Temporal patterns of wild boar-vehicle collisions in Estonia, at the northern limit of its range , 2016, European Journal of Wildlife Research.

[2]  X. Santos,et al.  Seasonal variation in vertebrate traffic casualties and its implications for mitigation measures , 2017 .

[3]  Xiaolin Hu,et al.  Traffic sign detection based on convolutional neural networks , 2013, The 2013 International Joint Conference on Neural Networks (IJCNN).

[4]  Clara Grilo,et al.  Do dry ledges reduce the barrier effect of roads , 2013 .

[5]  Sei-Wang Chen,et al.  Road-sign detection and tracking , 2003, IEEE Trans. Veh. Technol..

[6]  Torsten Hothorn,et al.  Temporal patterns of deer-vehicle collisions consistent with deer activity pattern and density increase but not general accident risk. , 2015, Accident; analysis and prevention.

[7]  Sinisa Segvic,et al.  A computer vision assisted geoinformation inventory for traffic infrastructure , 2010, 13th International IEEE Conference on Intelligent Transportation Systems.

[8]  Mohammed Taleb Obaidat,et al.  Integration of geographic information systems and computer vision systems for pavement distress classification , 2006 .

[9]  Majid Mirmehdi,et al.  Real-Time Detection and Recognition of Road Traffic Signs , 2012, IEEE Transactions on Intelligent Transportation Systems.

[10]  Filippo Sorbello,et al.  A neural network based automatic road signs recognizer , 2002, Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290).

[11]  Manuel Lagunas,et al.  Transfer Learning for Illustration Classification , 2018, CEIG.

[12]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[13]  Marcel P Huijser,et al.  Effectiveness of short sections of wildlife fencing and crossing structures along highways in reducing wildlife–vehicle collisions and providing safe crossing opportunities for large mammals , 2016 .

[14]  Luis Moreno,et al.  Road traffic sign detection and classification , 1997, IEEE Trans. Ind. Electron..

[15]  Eugenio Culurciello,et al.  An Analysis of Deep Neural Network Models for Practical Applications , 2016, ArXiv.