Vehicle Image Capture for Object Recognition Error Archiving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle-based image capturing systems for autonomous driving cannot be adequately checked for correct functioning during operation, leading to potential errors in object recognition, particularly with moving objects, which is crucial for safe autonomous driving.
Innovation Solution
A method for capturing image material that checks object recognition accuracy in terms of time and location, archiving only necessary images or image sections for subsequent analysis using odometry and trajectory data, with options for temporary storage in a vehicle or external archive.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all captured images are archived for subsequent analysis, then measurement precision of image analysis system is improved, but device complexity and memory requirements increase
Solution Approach 1:
The patent segments the image data archiving process by capturing only specific regions of interest (ROIs) rather than entire images. The computing device identifies objects in captured images, determines their locations, and archives only the image sections containing these objects along with metadata (position, time, speed). This segmentation approach maintains measurement precision for object recognition while significantly reducing memory requirements and device complexity.
Solution Approach 2:
The patent extracts and archives only the essential information needed for subsequent analysis: image sections containing identified objects, their position coordinates, timestamp, and vehicle speed. By taking out only the critical data elements rather than archiving complete images, the system achieves adequate measurement precision for verifying object recognition accuracy while minimizing memory consumption and system complexity.
2Reliability
If image material is archived for checking object recognition, then reliability of image analysis system is improved, but loss of time for data processing increases
Solution Approach 1:
The patent applies preliminary action by capturing and archiving image material in real-time during vehicle operation, rather than processing all data afterward. The computing device continuously captures images, identifies objects, and archives relevant image sections with metadata as they are detected. This preliminary archiving enables subsequent verification of object recognition accuracy without requiring extensive post-processing time, thus improving reliability while minimizing time loss.
Solution Approach 2:
The patent extracts only the necessary image sections and metadata (object position, time, speed) for archiving, excluding redundant data. This selective extraction reduces the volume of data requiring subsequent processing and analysis, thereby maintaining system reliability through adequate verification capability while significantly reducing the time lost to data processing operations.
3Measurement precision
If complete images are archived instead of image sections, then measurement precision is improved, but loss of substance (data volume) increases
Solution Approach 1:
The patent segments complete images into smaller image sections, archiving only those sections that contain identified objects along with their position coordinates. Instead of storing entire images, the system divides and stores only the relevant portions, maintaining sufficient measurement precision for verifying object recognition while dramatically reducing data storage volume. The segmentation approach preserves critical information while eliminating redundant background data.
Solution Approach 2:
The patent applies local quality by varying the archiving detail based on location significance: image sections containing identified objects are archived with higher detail and precision, while areas without objects are excluded. This local quality approach ensures that measurement precision is maintained where it matters most (at object locations) while reducing overall data storage requirements by omitting or reducing detail in non-critical areas.
Data Source
AI summary
Technologies and techniques for capturing image material for monitoring image-analyzing systems, wherein an object is monitored to determine if it has been correctly recognized by the image-analyzing system in respect of time or location. The images captured are recorded in a memory. When a discrepancy is determined, in object recognition beyond a tolerance limit over a temporal or locational reference, select images or image sections are archived for more precise inspection.