Bird's-Eye View Image Synthesis for Artifact-Free Segmentation Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for generating bird's eye view (BEV) images often result in distorted and artifact-laden images due to transformation processes, leading to inaccurate and difficult-to-process images that hinder the performance of driver assistance and vehicle automation features.
Innovation Solution
A system utilizing a machine learning component, specifically a generative adversarial model (GAN), is employed to generate synthesized BEV images and segmentation maps by removing artifacts from stitched images, thereby producing distortion-free and accurate images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a transformation process is used to generate BEV images from stitched images, then the BEV image can be generated, but the generated BEV image contains distortion and artifacts
Solution Approach 1:
A machine learning model acts as an intermediary between the stitched image and the final BEV image. The model first generates a preliminary BEV image through transformation, then processes this intermediate result to remove distortion and artifacts, producing a clean final BEV image. This two-stage approach with the ML model as mediator resolves the contradiction by separating the generation process from the quality refinement process.
Solution Approach 2:
The patent replaces traditional mechanical/image processing transformation methods with a machine learning-based approach. Instead of using fixed geometric transformation algorithms that inherently introduce distortion, the system uses a trained neural network that has learned optimal transformation patterns from data, substituting rigid mechanical transformation with adaptive intelligent processing to eliminate artifacts.
2Productivity
If existing transformation processes are used for BEV generation, then processing can be completed, but the images are difficult to process and reduce performance of driver assistance features
Solution Approach 1:
The machine learning model performs preliminary learning and training in advance to capture the optimal transformation patterns and characteristics of BEV images. During actual operation, the pre-trained model can quickly process stitched images into high-quality BEV images without requiring complex real-time calculations, thus maintaining high processing efficiency while ensuring reliable output quality for driver assistance systems.
Data Source
AI summary
Techniques are described for generating bird's eye view (BEV) images and segmentation maps. According to one or more embodiments, a system is provided comprising a processor that executes computer executable components stored in at least one memory, comprising a machine learning component that generates a synthesized bird's eye view image from a stitched image based on removing artifacts from the stitched image present from a transformation process. The system further comprising a generator that produces the synthesized bird's eye view image and a segmentation map, and a discriminator that predicts whether the synthesized bird's eye view image and the segmentation map are real or generated.