Automatic Recognition of Land Instability
S. Meziane1, L. Bahi2, L. Ouadif3
1MEZIANE Soukaina, Mohammadia School of Engineers, Mohammed V University, Rabat, Morocco.
2BAHI Lahcen, Mohammadia School of Engineers, Mohammed V University, Rabat, Morocco.
3OUADIF Latifa, Mohammadia School of Engineers, Mohammed V University, Rabat, Morocco.
Manuscript received on 01 March 2019 | Revised Manuscript received on 06 March 2019 | Manuscript published on 30 July 2019 | PP: 1972-1977 | Volume-8 Issue-2, July 2019 | Retrieval Number: B1967078219/19©BEIESP | DOI: 10.35940/ijrte.B1967.078219
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Land instabilities are frequent in vulnerable areas because of the conjunction of several triggers. Therefore, it is necessary to identify areas at risk and to propose preventive comfort solutions. The risk is getting worse if landslide affects road network. As a result, it reduces the security of users, the opening of areas and the access of the population. Automatic recognition of land instabilities is useful in the field of new mobility and smart infrastructures. Indeed, this recognition helps the autopilot system installed in vehicles to detect land instabilities even in in bad weather allowing the conditional autonomy to monitor the environment and then the vehicle is able to recognize the land instabilities and acts accordingly. Otherwise, it allows the development of intelligent and smart infrastructures for the protection of the platform and the support of slopes at risk. This automation will reduce the time needed to provide solutions and help diagnose potential risks at an early stage with the optimization of maintenance cost. This article proposes to apply a deep learning method for the recognition of land instabilities collected on roads from visual surveys or captured images of roads. Since landslide types are multiple and require special study to classify them distinctly, this paper aims to apply the deep convolutional neural network (DCNN) using pre-trained AlexNet model to distinguish two classes presence of land instability on the roadway and absence of land instabilities. The results of classification are satisfactory with saving time in identifying risk areas with excellent accuracy (85%) and efficiency and high F1-score (89.12%).
Index Terms: Deep Convolutional Neural Network, Roads, Landslides Recognition, Smart Infrastructures.
Scope of the Article: Pattern Recognition and Analysis