Neural Parking Spot Detection for Open-Space Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing radar-based automatic parking systems struggle to accurately detect parking spots in open spaces and often erroneously identify parking stoppers, leading to incomplete parking and poor positioning accuracy.
Innovation Solution
A method and system utilizing a neural network to process images of a parking lot, predicting center point coordinates and corner point probabilities, correcting these coordinates, and calculating final parking spot probabilities based on center and corner point probabilities, enhancing detection accuracy and reducing false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If radar-based automatic parking system is used, then parking spot between two vehicles can be detected, but detection of parking spot marked on open space fails and false detection of parking stopper occurs
Solution Approach 1:
The patent replaces the radar-based detection system with a vision-based system using cameras and neural network algorithms. This substitution enables the system to detect various parking spot types (marked spots on open spaces, spots between vehicles) by processing visual information through deep learning models, thereby improving both reliability and adaptability simultaneously
Solution Approach 2:
The patent changes the detection parameters from radar signals to visual features extracted from camera images. By using neural networks to analyze color, shape, and pattern characteristics of parking spots in images, the system can accurately distinguish between marked parking spots and parking stoppers, resolving the false detection issue while maintaining versatility
2Device complexity
If radar-based detection is used, then simple parking spot detection is achieved, but positioning accuracy deteriorates due to false detection of parking stoppers
Solution Approach 1:
The patent replaces the simple but inaccurate radar system with a vision-based system that uses neural networks for sophisticated image analysis. This allows the system to accurately distinguish parking spots from parking stoppers by recognizing visual patterns, thereby improving positioning accuracy without significantly increasing system complexity
Solution Approach 2:
The patent introduces neural network algorithms as an intermediary between the camera and the detection result. These algorithms process and interpret visual information, filtering out false detections of parking stoppers and accurately identifying true parking spots, thus improving measurement precision while maintaining system simplicity
Data Source
Figure 1~3
Figure 4~5
Figure 6
AI summary
The invention relates to a parking spot detection method and system. The method includes: acquiring a parking lot image including one or more idle and occupied parking spots around a vehicle; inputting the parking lot image to a neural network model for image processing, to obtain center point coordinates and a center point probability of the parking spot, first coordinates and first corner point probabilities of four corner points, and offsets of the four corner points relative to the center point coordinates; calculating second coordinates of the four corner points based on the center point coordinates and the offsets of the four corner points relative to the center point coordinates; correcting the second coordinates of the four corner points by using the first coordinates of the four corner points, to obtain actual coordinates and actual corner point probabilities of the four corner points; and performing parking spot detection based on the center point coordinates and the center point probability, and the actual coordinates and actual corner point probabilities of the four corner points. According to the invention, the parking spot can be accurately detected.