Sub-Pixel Ray Tracing for Scalable Ground Truth Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing real capture data for machine learning and computer vision tasks is less scalable and less accurate due to manual image capture and post-processing for ground truth generation, which can introduce inaccuracies.
Innovation Solution
A computer device simulates virtual environments using ray tracing to generate synthetics data with sub-pixel data, allowing for scalable and accurate image generation for training machine learning models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If real capture data is used for machine learning training, then ground truth data can be obtained through post-processing, but scalability is reduced due to manual capture and labeling requirements
Solution Approach 1:
The patent creates virtual copies of real-world environments through synthetic data generation. Instead of capturing real images manually, the system renders synthetic images that replicate real-world scenarios, eliminating the need for physical capture while maintaining data quality and enabling unlimited scalability.
Solution Approach 2:
The patent replaces the mechanical process of manual image capture and human labeling with an automated rendering system. The rendering engine automatically generates both images and ground truth data through computational processes, eliminating manual intervention and enabling high-speed data production.
2Productivity
If manual post-processing is used to generate ground truth data for real capture images, then labeled data can be created, but accuracy is reduced due to human labeling errors
Solution Approach 1:
The rendering system automatically generates ground truth data as an inherent part of the image creation process. The same rendering engine that produces images also computes accurate ground truth information (depth, segmentation, normals) without requiring separate manual labeling steps, ensuring consistency and high accuracy.
3Device complexity
If standard pixel-level data is used for image processing, then simplicity is maintained, but precision is lost due to lack of sub-pixel information
Solution Approach 1:
The patent segments each pixel into multiple sub-pixel components, where each sub-pixel represents a specific ray intersection property (depth, normal, segmentation label). This segmentation allows the system to maintain pixel-level data organization while capturing fine-grained sub-pixel information for improved processing accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computer device is provided that includes a processor configured to simulate a virtual environment based on a set of virtual environment parameters, and perform ray tracing to render a view of the simulated virtual environment. The ray tracing includes generating a plurality of rays for one or more pixels of the rendered view of the simulated virtual environment. The processor is further configured to determine sub-pixel data for each of the plurality of rays based on intersections between the plurality of rays and the simulated virtual environment, and store the determined sub-pixel data for each of the plurality of rays in an image file.