The goal of the project CrowDPLOS is to utilize computer vision in order to calculate a level of service for pedestrians with limited mobility such as disabled people. The project aims at extracting features from existing aerial images of the pedestrian network and from crowdsourced images of street segments through computer vision algorithms. Those features are then weighted and an accessibility score (i.e. the level of service) is provided for each street segment. These scores can be used afterwards to calculate routes adapted to the mobility level of the user or to inform urban planning initiatives. In this preliminary project we aim at:
- analyzing the possibilities of existing computer vision algorithms to automatically derive a Disabled Pedestrian Level of Service (DPLOS) for a pedestrian road graph from images;
- test and evaluate the identified algorithms using test-scenarios with manually defined DPLOS for each edge of the graph;
- evaluate the efficiency of crowdsourcing as a source of usable data by comparing it to existing aerial images;
- identify missing research gaps in this area and develop the research question of the SNF/FNS proposal.