This specification describes systems and methods for refining point cloud data. Methods can include receiving point cloud data for a physical space, iteratively selecting points along an x, y, and z dimension, clustering the selected points into 2D histograms, determining a slope value for each 2D histogram, and removing, based on the slope value exceeding a predetermined value, points from the point cloud data. Methods can also include iteratively voxelizing each 2D histogram into predetermined mesh sizes, summating points in each voxelized 2D histogram, removing, based on determining the summation is below a predetermined sum value, points from the point cloud data, keeping, based on determining that a number of points in each voxelized 2D histogram exceeds a threshold value, a center point, selecting, for each histogram, a point, identifying, nearest neighbors in the point cloud data, removing the identified nearest neighbors from the data, and returning remaining points.

A method and system of constructing a network map from imagery comprising using an iterative search process guided by a CNN-based decision function to derive a network graph directly from the output of the CNN.

Provided herein is a method for quantifying and measuring human mobility within defined geographic regions and sub-regions. Methods may include: identifying sub-regions within a region; identifying static information associated with the sub-regions from one or more static information sources; obtaining dynamic information associated with the sub-regions from one or more dynamic information sources; determining correlations between elements of the static information associated with a respective sub-region and elements of the dynamic information associated with the respective sub-regions; generating a mobility score for the respective sub-region based, at least in part, on the correlations between the elements of the static information and the elements of the dynamic information associated with the respective sub-region; and providing the mobility score to one or more clients for guiding an action relative to the mobility score.

A pothole monitoring and repair system for a road surface includes at least one UAV which includes a video camera, a video processor, a computing unit and a global positioning receiver. The UAV generates video streams of road surfaces using the video camera which are processed by the video processor to extract road frames. The computing unit generates a geo-fenced area of the road surface, and determines whether there is at least one pothole in each of the road frames within the geo-fenced area. The cloud server generates a map of the geo-fenced area, labels location coordinates of each pothole on the map, identifies and labels geographical features in the road frames within the geo-fenced area, extracts pothole features in the road frames, and uses a classifier to predict repair actions based on pothole features and geographical features, which are presented on a front end interface.

Systems and methods are described for identifying and validating the routes and characteristics of roads unknown to a road mapping database. The systems and methods may combine feature recognition analysis of aerial images with other information sources such as location tracking information from a mobile device or client in order to improve the accuracy of road information stored within a road mapping database. The systems and methods may also facilitate the collection of additional information regarding the characteristics of the identified roads from a client device or user thereof.

A system may be configured to collect geospatial features (in vector form) such that a software application is operable to edit an object represented by at least one vector. Some embodiments may: generate, via a trained machine learning model, a pixel map based on an aerial or satellite image; convert the pixel map into vector form; and store the vectors. This conversion may include a raster phase and a vector phase. A system may be configured to obtain another image, generate another pixel map based on the other image, convert the other pixel map into vector form, and compare the vectors to identify changes between the images. Some implementations may cause identification, based on a similarity with converted vectors, of a more trustworthy set of vectors for subsequent data source conflation.

A method of predicting one or more road segment attributes corresponding to a road segment in a geographical area, the method including: providing trajectory data and satellite image of the geographical area; calculating one or more image channels based on the trajectory data; and using at least one processor, classifying the road segment based on the one or more image channels and the satellite image using a trained classifier into prediction probabilities of the road attributes A data processing system including one or more processors configured to carry out a the method of predicting road attributes. A computer executable code including instructions for predicting one or more road segment attributes according to the method.

Embodiments are disclosed for ground plane estimation (GPE) using a LiDAR semantic network. In an embodiment, a method comprises: obtaining a point cloud from a depth sensor of a vehicle operating in an environment; encoding the point cloud; estimating, using a deep learning network with the encoded point cloud as input, a ground plane in the environment; planning a path through the environment based on a drivable area of the estimated ground plane; and operating the vehicle, the vehicle along the path. The deep learning network includes a two-dimensional (2D) convolutional backbone, a detection head for detecting objects and a GPE head for estimating the ground plane. In an embodiment, point pillars are used to encode the point cloud.

Systems and methods of automating the generation of a correction of an estimate of an elevation of a digital elevation model (DEM) of the bare earth under forest canopy. The disclosed embodiments facilitate generation of a more accurate DEM in areas of canopy coverage (where the input X-band DSM cannot see the ground) to estimate both the canopy height and the associated DEM. In some embodiments, the result of computationally correcting an estimate of an original DEM is a modified DEM. The method of correcting an estimate of an original DEM utilizes a pair of P-band radar images, an original DEM overlapping the same scene as the P-band radar images, at least one common, uniquely-identifiable point in the P-band radar images, and a definition of a geographical area surrounding the common, uniquely identifiable point over which the elevation correction is applicable.

An electronic device displays a representation of a field of view of a camera that includes a view of a three-dimensional space. The representation of the field of view is updated over time based on changes to current visual data detected by at least one of the one or more cameras. Movement of the electronic device moves the field of view of the camera in a first direction. While detecting the movement, the electronic device: updates the representation of the field of view in accordance with the movement; identifies one or more elements in the representation of the field of view that extend along the first direction; and, based at least in part on the determination of the one or more elements, displays, in the representation of the field of view, a guide that extends in the first direction and that corresponds to one of the identified elements.