Zebra Crossing Detection from Aerial Imagery Across Countries
We propose a data driven approach to detect zebra crossings in aerial imagery. The system automatically learns an appearance model from available geospatial data for an examined region. HOG as well as LBPH features, in combination with a SVM, yield state
- PDF / 1,224,853 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 38 Downloads / 200 Views
Abstract. We propose a data driven approach to detect zebra crossings in aerial imagery. The system automatically learns an appearance model from available geospatial data for an examined region. HOG as well as LBPH features, in combination with a SVM, yield state of the art detection results on different datasets. We also use this classifier across datasets obtained from different countries, to facilitate detections without requiring any additional geospatial data for that specific region. The approach is capable of searching for further, yet uncharted, zebra crossings in the data. Information gained from this work can be used to generate new zebra crossing databases or improve existing ones, which are especially useful in navigational assistance systems for visually impaired people. We show the usefulness of the proposed approach and plan to use this research as part of a larger guidance system. Keywords: Aerial imagery
1
· Street crossing · Visually impaired
Introduction
Visually impaired and blind people face a multitude of challenges in many aspects of their daily lives. Among those, outdoor as well as indoor navigation remains a great challenge due to the complexity of their surroundings and interactions with moving obstacles, such as other pedestrians or vehicles. Much needed visual cues are constantly utilized by unimpaired people, to avoid accidents and to not bump into one of the numerous obstacles, and are thus taken for granted. This effect is further amplified by an urban element design that neglects visually impaired peoples’ specific needs. Sidewalks are often not equipped with tactile pavings and items such as trash cans, lamp poles, or safety posts are often placed in spatially confusing or free space limiting locations, severely impeding movements by visually impaired people. Street crossings, as well as pedestrian traffic lights, often do not include much needed accessibility features and the general street layout can be very confusing to the visually impaired, e.g., Y-junctions, complicated round-abouts, or temporary construction sites. With respect to accessibility, zebra crossings are amongst the worst offenders, often providing no locational hints at all and only very limited tactile feedback while crossing the street. Furthermore, they provide less certainty and peace of mind compared to formal crossings [1]. c Springer International Publishing Switzerland 2016 K. Miesenberger et al. (Eds.): ICCHP 2016, Part II, LNCS 9759, pp. 27–34, 2016. DOI: 10.1007/978-3-319-41267-2 5
28
D. Koester et al.
Today’s ubiquitous Global Navigation Satellite Systems allow visually impaired people to regain at least some confidence in urban navigation situations. While these can greatly increase mobility, they only provide very high-level navigational information, but can neither warn of obstacles nor supply fine level guidance to the user. Thus, to this day, visually impaired people rely traditionally most on their white cane or a guide dog for these types of information. Furthermore, important geospatial informat
Data Loading...
