L. Kotian

A. Chheda

V. Narwane

R. Raut

Abstract

Unmanned Aerial Vehicle (UAV) can be used to monitor road conditions, particularly road survey and inspection of distressed roads. UAVs can collect large amounts of datasets and ease of use made these devices popular in many applications. On the basis of the detailed literature survey, various methods and models of UAV for road monitoring were studied. Lack of consideration of all distresses and inadequate data is the major research limitations. In this paper, a model for the preparation of an extensive collection of data and the classification of this data using semantic segmentation is proposed. The proposed model would efficiently help with huge data collection and the model designed will classify the distresses for maintenance and construction purposes. The model can be modified with neural networks or deep learning algorithms for accuracy and various frameworks can be used for improving speed and training time.

Keywords- Unmanned Aerial Vehicle (UAV), Potholes, Pavement distress, Fully Convolutional Networks (FCN), Semantic Segmentation