Object Recognition Using Wheel Detection for Pedestrian-Bicycle Distinction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image recognition systems for vehicles struggle to accurately distinguish between pedestrians and bicycles, especially when they are moving at similar speeds, and often incorrectly recognize bicycles or fail to detect them from diagonal or rearward angles, leading to inaccurate distance calculations and delayed warnings.
Innovation Solution
An object recognition apparatus that uses a combination of target and wheel recognition dictionaries to detect targets and wheels in images, with the wheel recognition dictionary providing a correction to the target detection, allowing for accurate identification and distance calculation, even when bicycles are partially obscured or viewed from unusual angles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single target recognition dictionary is used to detect all objects, then the device complexity is reduced, but the measurement precision of distinguishing between pedestrians and bicycles deteriorates
Solution Approach 1:
The patent divides the recognition system into two separate dictionaries: a target recognition dictionary for detecting the presence of objects and a wheel recognition dictionary for detecting wheel-specific features. This segmentation allows each dictionary to specialize in specific features, improving distinction between pedestrians and bicycles while maintaining manageable system complexity through modular architecture.
Solution Approach 2:
The patent adds a new dimension of recognition by introducing wheel detection as a separate feature layer. Instead of relying solely on target-level recognition, the system incorporates wheel presence and position as an additional dimensional cue, enabling more accurate differentiation between wheeled objects (bicycles) and non-wheeled objects (pedestrians).
2Measurement precision
If wheel recognition is added to improve target detection accuracy, then the measurement precision of target identification improves, but the device complexity increases
Solution Approach 1:
The recognition system is segmented into two independent but complementary dictionaries: target recognition dictionary and wheel recognition dictionary. Each dictionary processes specific features independently, allowing the system to achieve high detection accuracy through specialized feature detection while managing complexity through functional separation and modular processing.
Solution Approach 2:
The wheel recognition dictionary serves multiple functions: it detects the presence of wheels, determines wheel position relative to the target, and provides corrective information for target detection. This multi-functionality allows a single additional component to address multiple aspects of target identification, improving accuracy without proportionally increasing system complexity.
3Ease of operation
If the target recognition dictionary is used alone, then the ease of operation is maintained, but the reliability of distinguishing pedestrians from bicycles deteriorates
Solution Approach 1:
The wheel recognition dictionary provides feedback to the target recognition process by detecting wheel presence and position. This feedback mechanism allows the system to correct or confirm target identification results, significantly improving the reliability of distinguishing pedestrians from bicycles while maintaining ease of operation through automated multi-stage verification.
Solution Approach 2:
The recognition process is segmented into two sequential stages: target detection using the target recognition dictionary, followed by wheel verification using the wheel recognition dictionary. This segmentation enables reliable pedestrian-bicycle distinction through systematic feature verification while keeping the operation simple through automated multi-stage processing with clear functional separation.
Data Source
AI summary
An object recognition apparatus includes an image acquisition unit that acquires a captured image of a photographic subject, and a recognition processing unit that recognizes the photographic subject in the acquired image using a recognition dictionary. The recognition processing unit detects a target in the acquired image using a target recognition dictionary, detects a wheel at a lower part of the detected target using a wheel recognition dictionary, and reflects a result of the detection of the wheel in a result of the detection of the target. Thus, the object recognition apparatus can accurately detect a target such as a person and another vehicle.