Roundabout_Junctions_Dataset

At the ground level, a map is a detailed and accurate two-dimensional representation of natural and human- made features of the linear objects. These object can be in the form of a circle, rectangle or any combination of geometrical shapes. For domain analysis, attributes are used by merging the geometry and the real map. These attributes are defined by the practical study of the raw map data. One of the practical problems of maps related to the identification of junctions in a roundabout has been considered. This paper identifies the junctions in a roundabout, where the junctions can be entry or exit nodes, which refer to the point where the road enters into the roundabout or the point where the road exits from it. In this paper, this problem has been solved by using domain knowledge over real map data. Data from closed loops has been extracted which, in reality, are either roundabouts or not. On this data, multiple machine learning models have been implemented and domain analysis has been performed over the direction of travel that defines the traffic direction, which can be ‘BOTH’, ‘TO’ or ‘FROM’. Through this, it was possible to improvise the accuracy to 81 percent.