Abstract: Systems and methods for image-based self-localization of an image capture device, include receiving a query image of an undetermined location of the image and comparing the query image to a set of geotagged reference images to determine a location of the query image. The system and method may alternatively compare a series of query images to reference images or portions of feature vectors generated from said query images and reference images to determine the closest query image in the series to respective reference images. The set of geotagged reference images may include a sequence of previously obtained reference images of a route, each reference image corresponding to a known geolocation. The geolocation of the user may be determined based on the location of the query image within the set. Ambient conditions of the images may be used to improve comparison of a query image to reference images. Segment and/or abstractions of cell images may be used to reduce computational and/or communications resources.

Abstract: In order for the feature extractors to operate with sufficient accuracy, a high degree of training is required. In this situation, a neural network implementing the feature extractor may be trained by providing it with images having known correspondence. A 3D model of a city may be utilized in order to train a neural network for location detection. 3D models are sophisticated and allow manipulation of viewer perspective and ambient features such as day/night sky variations, weather variations, and occlusion placement. Various manipulations may be executed in order to generate vast numbers of image pairs having known correspondence despite having variations. These image pairs with known correspondence may be utilized to train the neural network to be able to generate feature maps from query images and identify correspondence between query image feature maps and reference feature maps. This training can be accomplished without requiring the capture of real images with known correspondence.