3D Data Generation from 2D Profile Annotations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual annotation of three-dimensional information for vehicles in automatic driving is expensive and time-consuming, while conventional two-dimensional annotation methods are insufficient for obtaining accurate three-dimensional data.
Innovation Solution
A method and apparatus that generate three-dimensional data by obtaining profile information annotation results from two-dimensional images, calculating depth and orientation of target objects, and determining three-dimensional coordinates, allowing for automatic generation of three-dimensional data using two-dimensional annotation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual three-dimensional annotation is used, then measurement precision of three-dimensional information is improved, but productivity and time consumption deteriorate
Solution Approach 1:
The patent uses two-dimensional annotation results as a copy or projection of the three-dimensional information, and through depth calculation and coordinate transformation, reconstructs the three-dimensional data from this two-dimensional copy, thereby avoiding direct manual three-dimensional annotation while maintaining accuracy
Solution Approach 2:
The patent introduces depth of field calculation and coordinate transformation as intermediary processes between two-dimensional annotation and three-dimensional data generation, using these intermediate mathematical operations to bridge the gap between 2D and 3D spaces
2Measurement precision
If manual three-dimensional annotation is used, then measurement precision of three-dimensional information is improved, but loss of time increases
Solution Approach 1:
The patent performs preliminary two-dimensional annotation which is faster and easier, then uses this preliminary result as the foundation for automated three-dimensional data generation through depth calculation and coordinate transformation, avoiding the need for time-consuming direct three-dimensional annotation
Solution Approach 2:
The method uses the two-dimensional annotation result as a copy that contains sufficient information to reconstruct three-dimensional data through mathematical transformations, significantly reducing the time required compared to direct manual three-dimensional annotation
3Productivity
If two-dimensional annotation method is used, then productivity is improved, but measurement precision of three-dimensional information deteriorates
Solution Approach 1:
The patent performs dimensionality transformation by taking two-dimensional annotation results and through depth of field calculation and coordinate transformation, reconstructing three-dimensional spatial information, thereby gaining three-dimensional measurement precision while maintaining two-dimensional annotation efficiency
Solution Approach 2:
The method changes the parameter representation from two-dimensional coordinates to three-dimensional coordinates through depth calculation and coordinate transformation, enabling accurate three-dimensional information extraction while maintaining the efficiency of two-dimensional annotation processes
4Measurement precision
If manual three-dimensional annotation is used, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent uses computationally efficient two-dimensional annotation as a copy that can be transformed into three-dimensional data through mathematical operations, avoiding the need for complex three-dimensional annotation tools and processes while maintaining accuracy
Data Source
AI summary
Embodiments of the present disclosure provide a method and an apparatus for generating three-dimensional data, a device, and a storage medium. The method comprises: obtaining a profile information annotation result for a target object in a two-dimensional image; calculating depth of field of the target object and a three-dimensional orientation of the target object according to the profile information annotation result and a standard three-dimensional size of the target object; determining a three-dimensional coordinate of a reference two-dimensional image point in the target object according to a two-dimensional coordinate of the reference two-dimensional image point in the two-dimensional image and the depth of field; and generating three-dimensional data matching the target object according to the three-dimensional coordinate of the reference two-dimensional image point and the three-dimensional orientation of the target object.


