Ecopia Tech Corporation

1000786269 Ontario Inc.

Yuanming Shu

Shuo Tan

Zihao Chen

Ruijie Deng

Dawei Zai

Hongbo Wang

Weiguang Ding

Tuan Zhao

Huansheng Mo

Andi Dai

Methods and systems for generating vector maps representing features depicted in imagery are provided. An example method involves obtaining remote imagery that depicts a feature, applying a machine learning model to the remote imagery to extract a geometric model of the feature encoded as a tokenized representation of a sequence of annotation operations, and interpreting the tokenized representation as a vector map representing the geometric model of the feature.

Machine learning for vector map generation

Methods and systems for determining the height of structures based on imagery of the structures and associated two-dimensional vector data are provided. An example method involves projecting two-dimensional vector data outlining a roof of a structure into images of the structure captured from different perspectives and feature matching the vector data across the imagery to determine a best-matching three-dimensional position for the roof situated at the height of the structure.

Vector data projection and feature matching to determine three-dimensional structure

Methods and systems for generating orthographic imagery are provided. An example method involves accessing a plurality of source images depicting a scene from multiple points of view, encoding at least some of the source images into corresponding series of multiscale feature maps, receiving a definition of an orthographic projection of the scene, and generating an orthographic image of the scene corresponding to the orthographic projection. The generating involves applying global attention across a set of higher-level features, applying a convolutional and upsampling layer, generating a depth map for the scene, and back-projecting each point on the depth map corresponding to a pixel that should be rendered back through the orthographic projection to determine a set of lower-level features to be used to decode the pixel, and applying local attention across the set of lower-level features to produce a set of decoded features from which pixel information can be determined.

Neural radiance fields for orthographic imagery

Methods and systems for generating artificial parcel data are provided. An example method involves accessing geospatial imagery that depicts at least one building and its surrounding area, applying at least one machine learning model to the geospatial imagery to generate artificial parcel data that represents a shape and location of a legal land parcel occupied by the building, accessing building footprint data comprising at least one polygon that represents a shape and location of the building, storing the artificial parcel data and the building footprint data in association with address data that represents an address of the building, receiving a query, the query comprising an address of the building or geospatial coordinates situated within the building footprint or artificial parcel data corresponding to the building, and providing, in response to the query, the building footprint data and the artificial parcel data corresponding to the building.

Machine learning for artificial parcel data generation

Methods and systems for generating geometric models of pitched roofs of buildings based on multiview imagery are provided. An example method involves establishing a three-dimensional coordinate space for the multiview imagery, displaying, through an annotation platform, multiview imagery depicting the pitched roof from different points of view, and providing, through the annotation platform, functionality for a user to define a perimeter of the pitched roof, define a ridge line of the pitched roof, and define at least part of a roof facet of the pitched roof by connecting the ridge line of the pitched roof to the perimeter of the pitched roof.

Annotation tools for reconstructing three-dimensional roof geometry

Methods and systems for enhancing elevation models with landcover feature data are provided. An example method involves accessing geospatial imagery covering an area of interest that includes a three-dimensional landcover feature, generating a digital surface model that includes one or more above-ground points corresponding to the three-dimensional landcover feature, extracting two-dimensional vector data from the geospatial imagery that represents a two-dimensional boundary of the three-dimensional landcover feature, generating a digital terrain model by removing from the digital surface model a portion that is bound by the two-dimensional vector data, generating three-dimensional vector data representing the three-dimensional landcover feature, inserting one or more ground-level points of the three-dimensional vector data into the digital terrain model, and filling the digital terrain model where the portion of the digital surface model was removed with estimated ground-level points of the three-dimensional vector data as a seed in a non-linear optimization process.

Enhancing elevation models with landcover feature data

Methods and systems for extracting and reviewing landcover feature data are provided. An example method involves accessing geospatial imagery covering a geographic area and landcover feature data extracted from the geospatial imagery, subdividing the geospatial imagery into a grid of image tiles, assigning landcover feature data review technicians to review the image tiles and associated landcover feature data through a landcover feature data review platform, and receiving revised landcover feature data that modifies the landcover feature data.

Platform for distributed landcover feature data review

Methods and systems for generating artificial parcel data are provided. An example method involves accessing geospatial imagery depicting one or more buildings and surrounding areas, applying a machine learning model to the geospatial imagery to generate artificial parcel data in a form of a distance-transform raster map that represents a legal land parcel for each of the one or more buildings, and converting the distance-transform raster map into a vector map containing one or more polygons that represent the boundaries of each of the legal land parcels.

Growing community of inventors

Toronto, Canada

Yuanming Shu