![](bga.png)
5 CONCLUSIONS
In this paper, we have presented the detailed design,
implementation and evaluation of a novel
road/terrain classification system. The proposed
system shows how mobile sensors can help to
automate and facilitate some of the more labor
intensive VGI tasks. Based on the analysis of
volunteered geographic information gathered by
bikers, geographic maps can be annotated
automatically with each of the 6 terrain types:
asphalt, cobblestones, tiles, gravel, grass, and mud.
In order to perform the terrain classification task,
the system employs a random decision forest (RDF),
fed with a set of discriminative image and
accelerometer features. The multi-sensor terrain
classification achieves 92% accuracy for the 6-class
terrain classification problem, and 97% accuracy for
the on-/off-road classification. Since the evaluation
is performed on data gathered during real bike runs,
these are ‘real-life’ accuracies.
Future work will focus on the influence of bike
conditions (e.g., speed and ascent/descent) on the
classification results. If someone is biking faster, for
example, it is expected that the accelerometer will be
more discriminative, while for slower bikers, visual
features will (probably) be. Further research is
needed to check these hypotheses and to incorporate
these kinds of dependencies in our system. Finally,
when no visual data is available, for example when
the camera is blocked or not facing the terrain, we
also think of using reverse geocoding techniques to
query and analyze online geographic data.
ACKNOWLEDGEMENTS
The research activities as described in this paper
were funded by Ghent University, iMinds,
University College West Flanders, the Institute for
the Promotion of Innovation by Science and
Technology in Flanders (IWT), the Fund for
Scientific Research-Flanders (FWO-Flanders), the
Belgian Federal Science Policy Office, and the EU.
REFERENCES
Bao, L. & Intille, S. S., 2004. Activity recognition from
user-annotated acceleration data, In Pervasive
Computing, Vol. 3001, pp. 1-17.
Breiman, L., 2001. Random forests. In Machine Learning,
Vol. 45, No. 1, pp. 5-32 (2001)
Ershad, S. F., 2011. Texture Classification Approach
Based on Combination of Edge & Co-occurrence and
Local Binary Pattern, In Proceedings of International
Conference on Image Processing, Computer Vision,
and Pattern Recognition, pp. 626-629.
Gall, J., Razavi, N. & Van Gool, L., 2012. An Introduction
to Random Forests for Multi-class Object Detection,
In LNCS Outdoor and Large-Scale Real-World Scene
Analysis, Vol. 7474, pp 243-263.
Goldman, J., Shilton, K., Burke, J., Estrin, D., Hansen, M.,
Ramanathan, N., Reddy, S., Samanta, V., Srivastava,
M. & West, R., 2009. Participatory Sensing - A
citizen-powered approach to illuminating the patterns
that shape our world. In White paper of Woodrow
Wilson International Center for Scholars – Foresight
and Governance project.
Haklay, M. & Weber, P., 2008. OpenStreetMap:User-
Generated Street Maps, In Pervasive Computing, Vol.
7, No. 4, pp. 12-18.
Khan, Y. N., Komma, P., Bohlmann, K. & Zell, A., 2011.
Grid-based Visual Terrain Classification for Outdoor
Robots using Local Features, In IEEE Symposium on
Computational Intelligence in Vehicles and
Transportation Systems (CIVTS 2011), pp. 16-22.
Khan, Y. N., Masselli, A. & Zell, A., 2012. Visual Terrain
Classification by Flying Robots, In IEEE International
Conference on Robotics and Automation (ICRA 2012),
pp. 498 – 503.
Kohavi R. & Provost, F., 1998. Glossary of terms, In
Machine Learning, Vol. 30, pp. 271-274.
Pietikäinen, M., Hadid, A., Zhao, G. &Ahonen, T., 2011.
Computer Vision Using Local Binary Patterns, In
Computational Imaging and Vision, Vol. 40.
Pinheiro, A. M. G., 2009. Image Descriptors Based on the
Edge Orientation, In Proceedings of the 4th
International Workshop on Semantic Media
Adaptation and Personalization, pp. 73-78.
Popescu, D., Dobrescu, R. & Merezeanu, D., 2008. Road
Analysis based on Texture Similarity Evaluation, In
Proceedings of the 7th WSEAS International
Conference on Signal Processing (SIP'08), pp. 47-51.
Ravi, N., Dandekar, N., Mysore, P. & Littman, M.L.,
2005. Activity Recognition from Accelerometer Data,
In Proceedings of the 17th Conference on Innovative
Applications of Artificial Intelligence, pp. 1541-1546.
Reddy, S., Shilton, K., Denisov, G., Cenizal, C., Estrin, D.
& Srivastava, M., 2010a. Biketastic: Sensing and
Mapping for Better Biking, In Proceedings of the
SIGCHI Conference on Human Factors in Computing
Systems (CHI’10), pp. 1817-1820.
Reddy, S., Mun, M., Burke, J., Estrin, D., Hansen M. &
Srivastava, M., 2010b. Using Mobile Phones to
Determine Transportation Modes, In Transactions on
Sensor Networks, Vol. 6, No. 2, pp. 13:1-27.
Srivastava, M., Abdelzaher, T. & Szymanski, B., 2012.
Human-centric Sensing, In Philosophical Transactions
of Royal Society, Vol. 370, No. 1958, pp. 176-197.
Shotton, J, Kim, T.-K. & Stenger, B., 2009. Boosting &
Randomized Forests for Visual Recognition (tutorial),
In International Conference on Computer Vision.
SIGMAP2013-InternationalConferenceonSignalProcessingandMultimediaApplications
48