Motor Vehicle Camera System Diagnosis via Dual Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current camera systems for motor vehicles lack a reliable method for diagnosing object recognition and classification algorithms, making it impossible to detect and correct potential detection or classification errors.
Innovation Solution
A diagnostic method that classifies detected objects using both sensor data and a classification model independently, allowing for the comparison of results to identify errors and prevent erroneous operations by generating a warning message or deactivating the camera system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If object detection algorithms and classification algorithms are used in a camera system, then object detection capability is improved, but the ability to check for detection errors and classification errors deteriorates
Solution Approach 1:
The patent segments the classification process into two independent parts: a first classification based on sensor data (rain, brightness, temperature) and a second classification based on image processing. By dividing the classification function, the system can compare results from both methods to detect errors, thus maintaining reliability while preserving detection capability.
Solution Approach 2:
The patent implements a feedback mechanism where the classification results from the first method (sensor-based) and second method (image-based) are compared. This comparison provides feedback to verify the accuracy of object detection and classification, enabling error detection without compromising the original detection algorithms.
2Measurement precision
If sensor data is used for classification, then classification accuracy under different environmental conditions is improved, but the complexity of the classification system increases
Solution Approach 1:
The patent divides the classification system into two segments: one that uses sensor data for environmental condition assessment and another that uses image processing. This segmentation allows the system to leverage sensor data for improved accuracy while maintaining a manageable structure by keeping the classification paths distinct and comparable.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for performing a diagnosis of a camera system (2) of a motor vehicle (1) by: providing at least one image (BD) by means of a camera (3); detecting an object (6) in the image (BD) by means of an image processing device; providing sensor data (SD) by means of at least one sensor (7) of the motor vehicle (1), wherein the sensor data (SD) characterizes environmental conditions of the motor vehicle (1); first classifying the object (6) and herein associating the object (6) with a class (K1, K2, K3, K4) among several predetermined classes (K1, K2, K3, K4) depending on the environmental conditions, wherein the classes (K1, K2, K3, K4) differ from each other with respect to the environmental conditions; second classifying the at least one object (6) and herein associating the object (6) with one of the classes (K1, K2, K3, K4) based on the image (BD) and independently of the sensor data (SD) by a classification device (12) using a predetermined classification model (11); and comparing classification results of the first and the second classification and performing a diagnosis depending on the comparison.