Parking Assist Vision for Region-Wide Parking Type Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing parking assisting devices struggle to accurately determine the type of parking (parallel, perpendicular, or angle) when images of license plates or wheels are unavailable, leading to incorrect evaluations.
Innovation Solution
A parking assisting device equipped with cameras and machine learning models to analyze vehicle positions and orientations, determining the most common parking type among surrounding vehicles based on vehicle body directions and travel angles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image recognition is performed based on license plates or wheels, then parking type evaluation can be achieved, but the system fails when images of these features are unavailable
Solution Approach 1:
The patent uses captured images of vehicles as alternative data sources when license plate or wheel images are unavailable. By analyzing vehicle body contours, positions, and orientations from these captured images, the system can still evaluate parking types without relying on specific vehicle features that may not be visible in all conditions.
Solution Approach 2:
The evaluation unit is designed to perform multiple evaluation methods: it can evaluate parking types based on license plates, wheels, or vehicle body positions/orientations. This multi-functional approach ensures the system can operate reliably regardless of which features are visible in the captured images.
2Loss of information
If parking type is evaluated for individual vehicles separately, then detailed information is obtained, but the overall parking type determination for a region becomes unreliable
Solution Approach 1:
The evaluation unit combines evaluation results from multiple individual vehicles to determine the overall parking type for a region. By aggregating data from several vehicles and identifying the most frequent parking orientation, the system achieves accurate region-wide parking type determination while preserving individual vehicle information.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
[Means for Resolution] A parking assisting device 100 according to the present embodiment comprises: an image acquiring portion 131 for acquiring a captured image PF around the vehicle 1; a vehicle detecting portion 133 for using the captured image PF to detect a plurality of other vehicles 2, parked around the vehicle 1; a type evaluating portion 134 for evaluating the parking type PK, indicating parallel parking PA, perpendicular parking PC, or angle parking PB, for each of the plurality of other vehicles 2; and a type determining portion 135 for determining, as the parking type PK for a region in which a plurality of other vehicles 2 are parked, the parking type PK with the greatest number of vehicles, of the plurality of other vehicles 2, based on the evaluation result by the type evaluating portion 134.