Back to results
US 12,586,297 B2Granted
Interactive image generation
AssigneeOutward, Inc.
InventorsClarence Chui, Manu Parmar, Akshansh Chaudhry, Himanshu Jain, Andrew Pang, Lakshay Sharma, Brook Aaron Seaton
Life of this patent
Estimating the term…
Abstract
A content generation platform is generally described herein. More specifically, interactive image generation and techniques and features thereof are disclosed herein. An arbitrary input image captured in an unknown environment is processed using one or more machine learning based networks to generate an interactive reconstruction or floorplan of the input image.
Claims54
A method, comprising:
- receiving a single input image of a scene captured in an unknown environment, wherein the single input image comprises an arbitrary image with undefined characteristics;
- processing the received single input image of the scene using one or more machine learning based networks, wherein at least one of the one or more machine learning based networks is trained at least in part on training images generated from three-dimensional models and constrained to a prescribed scene type to which the scene belongs;
- and
- generating an interactive reconstruction of the scene based on processing the single input image of the scene using the one or more machine learning based networks, wherein the generated interactive reconstruction of the scene comprises a two-dimensional image with at least partial three-dimensional capabilities but without having an underlying three-dimensional model, wherein the generated interactive reconstruction of the scene comprises a plurality of interactive features modifiable by an end user according to user preferences, and wherein the at least partial three-dimensional capabilities are facilitated by the one or more machine learning based networks having learned interactive features from three-dimensional models of training images.
1 more element
A system, comprising:
- a processor configured to:
- receive a single input image of a scene captured in an unknown environment, wherein the single input image comprises an arbitrary image with undefined characteristics;
- process the received single input image of the scene using one or more machine learning based networks, wherein at least one of the one or more machine learning based networks is trained at least in part on training images generated from three-dimensional models and constrained to a prescribed scene type to which the scene belongs;
- and
- generate an interactive reconstruction of the scene based on processing the single input image of the scene using the one or more machine learning based networks, wherein the generated interactive reconstruction of the scene comprises a two-dimensional image with at least partial three-dimensional capabilities but without having an underlying three-dimensional model, wherein the generated interactive reconstruction of the scene comprises a plurality of interactive features modifiable by an end user according to user preferences, and wherein the at least partial three-dimensional capabilities are facilitated by the one or more machine learning based networks having learned interactive features from three-dimensional models of training images;
- and
- a memory coupled to the processor and configured to provide the processor with instructions.
4 more elements
A computer program product embodied in a non-transitory computer readable medium and comprising computer instructions for:
- receiving a single input image of a scene captured in an unknown environment, wherein the single input image comprises an arbitrary image with undefined characteristics;
- processing the received single input image of the scene using one or more machine learning based networks, wherein at least one of the one or more machine learning based networks is trained at least in part on training images generated from three-dimensional models and constrained to a prescribed scene type to which the scene belongs;
- and
- generating an interactive reconstruction of the scene based on processing the single input image of the scene using the one or more machine learning based networks, wherein the generated interactive reconstruction of the scene comprises a two-dimensional image with at least partial three-dimensional capabilities but without having an underlying three-dimensional model, wherein the generated interactive reconstruction of the scene comprises a plurality of interactive features modifiable by an end user according to user preferences, and wherein the at least partial three-dimensional capabilities are facilitated by the one or more machine learning based networks having learned interactive features from three-dimensional models of training images.
1 more element
Dependencies are inferred from the claim text.