A View from the Track: Measuring Spatial Quality of Slow Mobility Routes. Possible Integration of GIS and Machine Learning Based Methods

Abstract

The specific aim of this research is to assess the spatial quality of a slow mobility route, integrating two sources of information: the first one based on a quantitative method and the second one focused on a more qualitative oriented one. The first one is a GIS-based analysis which corresponds to a 'looking from above', based on a geographic approach. The second one is based on the 'street level' approach, that is more perceptive, by using Machine Learning techniques based on Google Street View images. The distinction of this work from similar research is related to its focus on the outdoors. The Spatial Quality Index of Slow Routes (SQISR) method allows to compute an index on the base of a set of heterogeneous indicators, by using GIS mapping techniques. The second method is a peculiar application of Image-Recognition technology based on Machine Learning tools, which analyses web images. This methodology produces a set of words, which characterizes the spatial components of the landscape crossed by the route, by arranging the tags into four semantic fields: urban, outdoors, road and car. The applied methodologies give heterogeneous results when valuable elements are far from route. Results correlate in areas where natural elements are present. Further evolution of this research will be based on Machine Learning Image Library customization and improved code-based data acquiring methodology.