# US 11,908,185 B2 · Roads and grading detection using satellite or aerial imagery > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,908,185 B2 - Document ID: US11908185B2 - Status: Granted - Kind: B2 - Application number: 17/854,294 - Filed: 2022-06-30 - Published: 2024-01-04 - Granted: 2024-02-20 - Earliest claimed: 2022-06-30 - Assignees: Metrostudy Inc. - Applicants: Corentin Guillo; Sivakumaran Somasundaram - Inventors: Corentin Guillo; Sivakumaran Somasundaram - Web page: https://patentgrep.com/patent/US11908185B2 ## CPC - G06V 20/182: {Network patterns, e.g. roads or rivers} - G06V 10/40: Extraction of image or video features - G06V 10/454: {Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]} - G06V 10/82: using neural networks - G06V 20/13: Satellite images - G06V 20/176: {Urban or other man-made structures} - G06V 20/70: Labelling scene content, e.g. deriving syntactic or semantic representations - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract Methods, non-transitory computer-readable storage media, and computer or computer systems directed to detecting, analyzing, and tracking roads and grading activity using satellite or aerial imagery in combination with a machine learned model are described. ## Claims ### Claim 1 A method comprising: receiving a satellite or aerial image, or portion thereof, of a geographic area as input into a machine learned model trained with a set of satellite or aerial images capturing roads and grading which are each labeled in the set of images; predicting which features in the satellite or aerial image or portion thereof correspond to roads and which features correspond to grading by way of the machine learned model; and sending an output from the machine learned model comprising the predictions. ### Claim 2 (depends on 1) The method of claim 1, wherein the machine learned model is a trained Convolutional Neural Network (CNN). ### Claim 3 (depends on 1) The method of claim 1, wherein the output comprises an image having a grid dividing the image into sections and a prediction for a feature represented in each section. ### Claim 4 (depends on 1) The method of claim 1, wherein the machine learned model is stored on at least one non-transitory computer-readable memory. ### Claim 5 (depends on 1) The method of claim 1, wherein the receiving, predicting, and sending are performed by one or more processor. ### Claim 6 (depends on 1) The method of claim 1, further comprising storing the satellite or aerial image, or portion thereof, of a geographic area on at least one non-transitory computer-readable memory. ### Claim 7 (depends on 1) The method of claim 1, wherein the satellite or aerial image or portion thereof is received from a satellite or aerial imagery provider service. ### Claim 8 (depends on 1) The method of claim 1, wherein the output comprises a probability for each prediction that the prediction is correct. ### Claim 9 (depends on 1) The method of claim 1, further comprising converting pixel locations in the satellite or aerial image or portion thereof to geographic coordinates, and outputting the geographic coordinates. ### Claim 10 (depends on 1) The method of claim 1, wherein the output is displayed on the satellite or aerial image or portion thereof provided as input to the machine learned model. ### Claim 11 (depends on 1) The method of claim 1, wherein the features corresponding to roads and the features corresponding to grading are edges or geometric patterns within pixel data in the satellite or aerial image or portion thereof. ### Claim 12 (depends on 1) The method of claim 1, wherein the roads and grading are features of one or more active new residential housing development construction projects. ### Claim 13 (depends on 1) The method of claim 1, wherein the image and set of images are satellite images. ### Claim 14 (depends on 1) The method of claim 1, wherein the output comprises text comprising: a feature name; a label for the feature representing a prediction chosen from road, grading, and none; a decimal probability that the label for the feature is correct; and a centroid and polygonal boundary for the feature, both expressed as geographical coordinates; wherein the text is formatted and standardized for input into a Geographic Information System application. ### Claim 15 One or more non-transitory, computer-readable storage media having instructions for execution by the one or more processors, the instructions programmed to cause the one or more processors to: receive a satellite or aerial image, or portion thereof, of a geographic area as input into a machine learned model trained with a set of satellite or aerial images capturing roads and grading which are each labelled in the set of images; predict which features in the satellite or aerial image or portion thereof correspond to roads and which features correspond to grading by way of the machine learned model; and send an output from the machine learned model comprising the predictions. ### Claim 16 (depends on 15) The one or more non-transitory, computer readable storage media of claim 15, wherein the machine learned model is a trained Convolutional Neural Network (CNN) stored thereon. ### Claim 17 (depends on 15) The one or more non-transitory, computer readable storage media of claim 15, wherein the output comprises an image having a grid dividing the image into sections and a prediction for a feature represented in each section. ### Claim 18 A computer or computer system, comprising: one or more processors designed to execute instructions; one or more non-transitory, computer-readable memories storing program instructions for execution by the one or more processors, the instructions programmed to cause the one or more processors to: receive a satellite or aerial image, or portion thereof, of a geographic area as input into a machine learned model trained with a set of satellite or aerial images capturing roads and grading which are each labeled in the set of images; predict which features in the satellite or aerial image or portion thereof correspond to roads and which features correspond to grading by way of the machine learned model; and send an output from the machine learned model comprising the predictions. ### Claim 19 (depends on 18) The computer or computer system of claim 18, wherein the machine learned model is a trained Convolutional Neural Network (CNN) stored on one or more of the non-transitory computer readable memories. ### Claim 20 (depends on 18) The computer or computer system of claim 18, wherein the output comprises an image having a grid dividing the image into sections and a prediction for a feature represented in each section.