Class-Specific Likelihood Thresholds for Vehicle Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object recognition systems for vehicles do not efficiently differentiate between object classes, leading to inconsistent detection thresholds that can result in either missed objects or unnecessary vehicle operations, particularly concerning safety risks like pedestrian detection.
Innovation Solution
An object recognition apparatus that uses a storage device to store tolerance information for each object class, allowing a processor to calculate class-specific likelihood thresholds based on detected objects and environmental conditions, such as self-position and peripheral environment conditions like rainfall and illuminance, to optimize detection efficiency and safety.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the likelihood threshold is set low to reduce the possibility of undetected objects, then the detection coverage is improved, but the possibility of erroneous detection increases
Solution Approach 1:
The patent applies different likelihood thresholds for different object classes (pedestrian, vehicle, animal, etc.). Each class has its own threshold value stored in the storage device, allowing the system to optimize detection parameters locally for each object type rather than using a single global threshold. This resolves the contradiction by enabling high detection coverage for critical objects like pedestrians while maintaining accuracy for other classes.
Solution Approach 2:
The system dynamically changes the likelihood threshold parameter based on the detected object class. When an object is identified as a pedestrian, a lower threshold is applied to ensure detection; when identified as a stationary object, a higher threshold is applied to reduce false positives. This parameter adaptation resolves the contradiction between coverage and accuracy.
2Measurement precision
If the likelihood threshold is set high to reduce erroneous detection, then the detection accuracy is improved, but the possibility of undetected objects increases
Solution Approach 1:
Different object classes are assigned different threshold values to balance accuracy and coverage locally. Critical moving objects like pedestrians use lower thresholds for higher coverage, while stationary objects use higher thresholds for higher accuracy. This local differentiation resolves the contradiction.
Solution Approach 2:
The likelihood threshold is not fixed but dynamically selected based on the object class identification. The system adapts the threshold parameter in real-time according to what type of object is detected, making the detection system flexible and context-aware to resolve the accuracy-coverage trade-off.
3Device complexity
If a single likelihood threshold is used for all object classes, then the system complexity is reduced, but the detection efficiency for different object classes deteriorates
Solution Approach 1:
The detection system segments object classes into distinct categories (pedestrian, vehicle, animal, etc.), each with its own likelihood threshold stored in the storage device. This segmentation allows optimized detection parameters for each class while maintaining a relatively simple system structure using a lookup table approach.
Solution Approach 2:
The storage device stores multiple threshold values that can be universally applied to different object classes. The same basic detection algorithm and threshold selection mechanism handles all object types, providing a universal solution that improves efficiency across different classes without requiring completely separate detection systems.
Data Source
AI summary
An object recognition apparatus include a storage device and a processor. The storage device stores peripheral information and tolerance information. The tolerance information is information in which the degree of tolerance for the undetected object is represented for each class of the object. The peripheral information is acquired by a sensor device provided in the vehicle. The processor performs object recognition process for recognizing an object around the vehicle. In the object recognition process, the processor identifies the object and its class to be detected based on the peripheral information, and calculates the likelihood that is a parameter representing the probability of detection of the object. Further, the processor calculates a likelihood threshold corresponding to the object based on the tolerance information, and determines whether to output the identification result of the object based on the comparative between the likelihood and the likelihood threshold.