Automatic Discovery and Geotagging of Objects from Street View Imagery

Abstract

Abstract:Many applications, such as autonomous navigation, urban planning, and asset monitoring, rely on the availability of accurate information about objects and their geolocations. In this paper, we propose the automatic detection and computation of the coordinates of recurring stationary objects of interest using street view imagery. Our processing pipeline relies on two fully convolutional neural networks: the first segments objects in the images, while the second estimates their distance from the camera. To geolocate all the detected objects coherently we propose a novel custom Markov random field model to estimate the objects' geolocation. The novelty of the resulting pipeline is the combined use of monocular depth estimation and triangulation to enable automatic mapping of complex scenes with the simultaneous presence of multiple, visually similar objects of interest. We validate experimentally the effectiveness of our approach on two object classes: traffic lights and telegraph poles. The experiments report high object recall rates and position precision of approximately 2 m, which is approaching the precision of single-frequency GPS receivers.

Automatic detection of passable roads after floods in remote sensed and social media data

“…In recent years, several applications have emerged to make use of data posted on social media platforms in combination to other media streams available (e.g. Google Street view imagery, OpenStreetMap map) allowing 3D reconstruction of cities [46] and automatic discovery and geotagging of objects [34].…”

Simultaneous Multi-View Instance Detection With Learned Geometric Soft-Constraints

“…Object geo-localization: Geo-localization of objects from Google street-view imagery with noisy pose information was introduced in [30,3]. In a similar attempt, [14] geo-localize traffic lights and telegraph poles by applying monocular depth estimation using CNNs, then using a Markov Random Field model to perform object triangulation. The same authors extend their approach by adding LiDAR data for object segmentation, triangulation, and monocular depth estimation for traffic lights [15].…”

Fully Convolutional Networks for Street Furniture Identification in Panorama Images

“…For 2D images such as street views and color images which are captured by cameras mounted on vehicles, image segmentation is regard as one of the most essential tasks for object extraction. A number of methods have been developed to solve this problem, from elementary pattern analysis like Hough transformation, via feature extraction-based tools like boosting, to more advanced machine learning and deep learning algorithms such as Support Vector Machines, Conditional Random Field, and Convolutional Neural Networks (Krylov et al 2018). To detect specific objects effectively, researchers proposed their new machine learning descriptors or improve the results by integrating with someone else's machine learning model, such as to extract utility poles (Zhang et al, 2018) by RetinaNet object detector (Lin et al, 2017).…”